Repositorio Institucional de Documentos

2015 1 Carlos Escolano Marco A subject-specific neurofeedback approach for cognitive enhancement Departamento Director/es Informática e Ingeniería de Sistemas Mínguez Zafra, Javier Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Carlos Escolano Marco A SUBJECT-SPECIFIC NEUROFEEDBACK APPROACH FOR COGNITIVE ENHANCEMENT Director/es Informática e Ingeniería de Sistemas Mínguez Zafra, Javier Tesis Doctoral Autor 2014 Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA A subject-specific neurofeedback approach for cognitive enhancement Carlos Escolano Marco Thesis submitted in fulfillment of the requirements for the degree of Doctor of Philosophy in Computer Science and Systems Engineering Supervised by Javier Mínguez Instituto de Investigación en Ingeniería de Aragón (I3A) Departamento de Informática e Ingeniería de Sistemas (DIIS) Escuela de Ingeniería y Arquitectura (EINA) Universidad de Zaragoza October, 2014 ii Resumen La técnica de neurofeedback (NF) permite el aprendizaje de la autoregulación de la actividad cerebral, por la cual los usuarios pueden aprender a modificar (hasta un cierto grado) patrones de la actividad cerebral. Un punto clave del NF en la práctica es la técnica de registro de la actividad cerebral, donde el electroencefalograma (EEG) es la más usada debido a que es no invasiva, portátil y tiene una buena resolución temporal. La investigación en neurociencia ha reportado en repetidas ocasiones la relación de determinadas funciones cognitivas y desórdenes psiquiátricos con las oscilaciones del EEG. Por lo tanto, no es sorprendente que la regulación de estas oscilaciones conlleve efectos en comportamiento. Por ejemplo, muchos estudios han aplicado esta técnica de NF para aumentar funciones cognitivas como memoria de trabajo, atención y habilidades visuoespaciales, usualmente evaluadas en usuarios sanos. Además se ha aplicado NF para el tratamiento de desórdenes psiquiátricos tales como el trastorno por déficit de atención con hiperactividad (TDAH), depresión, epilepsia y tinitus, entre otros. El EEG es una señal no estacionaria que presenta una variabilidad inherente entre usuarios en los correlatos de procesos mentales. Sin embargo, la gran mayoría de técnicas de NF desarrolladas hasta el momento son genéricas en el sentido de que no se adaptan a los patrones individuales de EEG de cada usuario. En este sentido ha habido una tendencia en los últimos años hacia el uso de técnicas de NF específicas por medio de la adaptación de algunos métodos involucrados en el procedimiento. Esta tesis aborda el diseño de una técnica de NF dentro de un framework unificado, y muestra su viabilidad por medio de la implementación de tres estudios de NF consistentes en el incremento de la actividad en alpha superior para mejora cognitiva, evaluada en usuarios sanos, pacientes con depresión mayor y niños con TDAH. Una disciplina que viene a la mente cuando pensamos en cómo individualizar la técnica de NF son las interfaces cerebro-computador (BCIs). Las BCIs es una tecnología reciente cuyo objetivo es abrir un canal de comunicación entre un humano y un dispositivo usando únicamente la actividad cerebral para mejor la calidad de vida de personas con graves deficiencias moiii iv toras. Un punto clave de las BCIs es que los métodos de procesamiento de la señal se adaptan a cada usuario y momento de utilización de la tecnología. Esto se logra usualmente por medio de una fase de calibración ejecutada antes de la operación en línea. En esta fase de calibración los usuarios realizan tareas mentales que permiten medir algunos correlatos de EEG que son usados para calcular filtros individualizados, por ejemplo en términos de filtrado de artefactos y detección de tareas mentales. Después estos filtros se aplican en línea al EEG para decodificar las tareas mentales, accionando en consecuencia el dispositivo. Tomando el framework de BCI como referencia, proponemos los siguientes métodos individualizados: (1) un método de filtrado de artefactos en tiempo real (usando separación ciega de fuentes) para eliminar los artefactos de los patrones cerebrales de interés; (2) un método novedoso para la individualización de los patrones cerebrales de acuerdo a la combinación de registros de EEG en dos condiciones (estado de reposo y tarea activa); (3) un método para calcular el nivel de trabajo de los patrones cerebrales (baseline) por sujeto y sesión; y (4) una variedad de métodos y métricas para evaluar los efectos del NF en los patrones cerebrales (post-análisis). Para evaluar la viabilidad y validez experimental de esta técnica de NF, llevamos a cabo una implementación de un protocolo de NF basado en el incremento de la actividad en alpha superior para mejora cognitiva, evaluada en tres estudios distintos de NF involucrando usuarios sanos, pacientes con depresión y niños con TDAH. Estos estudios investigaron si individuos sanos, con depresión y ADHD son capaces de incrementar la potencia en alpha superior por medio de NF y, si es así, si estos efectos están relacionados con efectos en comportamiento (rendimiento cognitivo o escalas clínicas). 1. El primer estudio investigó los efectos de una única sesión de NF en usuarios sanos (N = 19) en un diseño experimental con falso feedback. Este estudio mostró un incremento en la potencia en alpha superior en la tarea activa (inmediatamente después del NF) así como un incremento en potencia en alpha superior durante la tarea de rotación mental (intervalo pre-estímulo), únicamente para el grupo experimental. Ambos grupos mejoraron en rendimiento cognitivo, con una mejora superior para el grupo experimental. Sin embargo una única sesión parece insuficiente para producir diferencias significativas entre grupos. 2. El segundo estudio investigó los efectos de ocho sesiones de NF en pacientes con depresión (N = 60) en un estudio controlado. Este estudio mostró un incremento en la potencia en alpha superior en la tarea activa (pre-post estudio) para el grupo experimental. Estos efectos no v estuvieron restringidos espacialmente o espectralmente al parámetro de NF. Se encontró un incremento de actividad a nivel de las fuentes cerebrales en alpha para el grupo experimental, localizado en el giro cingulado anterior (sgACC, BA 25). El grupo experimental mostró un incremento en rendimiento así como un incremento en velocidad de procesamiento medido por un test de memoria de trabajo después del NF, sugiriendo por tanto que los síntomas cognitivos de pacientes con depresión pueden aliviarse por medio de este procedimiento. 3. El tercer estudio investigó los efectos de 18 sesiones de NF en niños con TDAH (N = 20) en un estudio preliminar no controlado. Este estudio mostró un incremento en la potencia (relativa y absoluta) en alpha superior en la tarea activa (pre-post estudio). Mientras que los cambios pre-post estudio estuvieron restringidos mayormente a la banda alpha superior, los efectos dentro de la sesión mostraron un decremento en potencia absoluta en las bajas frecuencias (se debe notar que los niños con TDAH usualmente tienen un exceso de actividad en bajas frecuencias). Los padres indicaron una mejora clínica en los niños con respecto a inatención e hiperactividad/impulsividad, y los tests neuropsicológicos mostraron una mejora en memoria de trabajo. En resumen, estos resultados muestran que la técnica de NF se adaptó a la gran variabilidad de los patrones cerebrales entre sujetos y sesiones. Además, estas tres poblaciones fueron capaces de auto-regular los patrones cerebrales con una consecuente mejora tanto en rendimiento cognitivo y como en escalas clínicas (en el caso de niños con TDAH). Aunque la principal contribución de esta tesis está en los métodos y en la implementación de una técnica individualizada de NF, los estudios de NF aquí presentados son novedosos en sí mismos y los resultados que se extraen de ellos constituyen una contribución añadida de esta tesis. vi Abbreviations ACC Anterior cingulate cortex ADHD Attention-deficit/hyperactivity disorder BCI Brain-computer interface EEG Electroencephalography fMRI Functional magnetic resonance imaging IAF Individual alpha frequency MDD Major depressive disorder NF Neurofeedback rTMS Repetitive transcranial magnetic stimulation SMR Sensorimotor rhythm UA Upper alpha WM Working memory xiii xiv Contents 1|Introduction No feedback, no learning. Feedback is the return of information an entity gets as a result of its actions, and it is a key factor to promote learning through operant conditioning (Skinner, 1938). Behavior can be reinforced or inhibited by means of positive and/or negative feedback evaluating the actual actions. These principles can be followed to promote the learning of the self-regulation of brain activity within a paradigm called neurofeedback (NF), where subjects can learn to shape (to a certain degree) some patterns of brain activity. The rationale of NF is to record the brain activity, to process the signal to decode the brain patterns of interest, and to provide the subjects with real-time feedback covarying with the brain patterns. A key practical point is the recording technique of brain activity, where the electroencephalogram (EEG) is the most widely used one as it is non-invasive, portable and presents a good temporal resolution. Neuroscience research has repeatedly reported the relation of cognitive functions and some psychiatric disorders to EEG oscillations (Başar and Güntekin, 2008, Klimesch et al., 2007, Klimesch, 1999). Thus, it is not surprising that the regulation of EEG oscillations yields behavioral effects. For instance, a large body of research has applied NF for the enhancement of cognitive functions such as working memory, attention and visuospatial abilities, usually applied to healthy subjects (Gruzelier, 2013, Vernon, 2005). NF has been also applied for the treatment of psychiatric disorders such as attention-deficit/hyperactive disorder (ADHD), depression, epilepsy and tinnitus, among others (Niv, 2013, Gruzelier, 2013). EEG is a non-stationary signal that presents an inherent variability among subjects in the EEG correlates of brain processes (Niedermeyer and da Silva, 2005). However, the large majority of NF procedures developed to date are subject-generic in the sense that they are not adapted to the individual EEG patterns of each subject. In this direction, there has been a trend in recent years towards the use of subjectspecific NF procedures by adapting (individualizing) some methods involved in this procedure (Hanslmayr et al., 2005, Zoefel et al., 2011, Nan et al., 1 2Chapter 1. Introduction EEG SIGNAL DECODER FEEDBACK BRAIN COMMAND LOW-LEVEL CONTROLLER EEG SIGNAL COMPUTE FILTERS ONLINECALIBRATION Figure 1.1: BCI for controlling of a robotic device. First, in the calibration phase a filter is computed according to the features of the subject’s EEG signal. Subsequently, this filter is used during online operation to decode the ongoing EEG signal and finally the subject receives feedback (visual in this example) as the actions executed by the device. 2012, Alexeeva et al., 2012). This thesis addresses the design of a subjectspecific NF approach in a methodological unified framework, and shows its feasibility by implementing three different NF studies of upper alpha upregulation for cognitive enhancement in healthy subjects, patients with major depressive disorder and children diagnosed with ADHD. One discipline that comes to mind when thinking about how to individualize EEG-based NF procedures is the brain-computer interfaces (BCIs). BCIs is a recent technology whose objective is to open a communication channel between a human and a device using only brain activity (Wolpaw et al., 2002) to improve the quality of life of people with severe motor disability (Millán et al., 2010). Some applications using the EEG as the recording technique include a speller for communication (Birbaumer et al., 1999), robotic devices for motor substitution (Iturrate et al., 2009) and neuroprosthetic devices for motor rehabilitation (Ramos-Murguialday et al., 2013). A key point of BCIs is that the signal processing methods are adapted for each subject and time of use of the technology. This is commonly achieved by a calibration phase before the online operation. In this calibration phase, the subjects perform metal tasks that allow to measure some EEG correlates that are used to compute subject-specific filters, for example in terms of filtering the EEG artifacts and detecting the mental tasks. These filters are then applied during the online operation phase to the ongoing EEG to decode the mental tasks, actuating the devices accordingly (Figure 1.1). For instance, in a BCI application for controlling a robot, the system initially 3 trains a classifier to detect the P300 event-related potential in the ongoing EEG, whose appearance is associated with a location of the space. During online operation the subject focuses on a location and a visual stimulation process elicits the P300 potential that enables the classifier to detect the subject’s desired location to navigate to, finally sending the relevant movement commands to the robot (Escolano et al., 2012a)1. Note that this calibration process is carried out every time the BCI is used, thus the outcome of this phase are specific filters for each subject and session. Taking the BCI framework as a reference, there are some issues that a subject-specific NF procedure might take into account: 1. EEG artifacts such as eye blinking and muscular artifacts may interfere with the brain patterns of interest (Delorme et al., 2007). The large majority of these artifacts are dependent on each subject and time of use of the technology (session in the context of NF). 2. EEG brain patterns present an inherent inter-subject variability (Haegens et al., 2014, Neuper et al., 2005). This inter-subject variability might be even higher in some clinical populations such as ADHD (Barry et al., 2003, Lansbergen et al., 2011b). 3. EEG brain patterns present an inherent inter-session variability within each subject (Haegens et al., 2014, Angelakis et al., 2004). This variability could be due to the non-stationary nature of the EEG (Niedermeyer and da Silva, 2005), to the physiological state of the subjects and the task being performed (Klimesch, 1999), or could be a consequence of the self-regulation (through NF) of these patterns. 4. As a consequence of addressing the three previous points with new signal processing methods, post-analysis methods that aggregate the EEG results in a subject- and session-specific way should be developed. The objective of this thesis is to design a subject-specific NF approach addressing the previous issues in a unified framework. We propose the following individualized methods: (1) a real-time artifact filtering method (using blind source separation) to remove the artifacts from the brain patterns of 1In the initial stage of this thesis we developed a BCI telepresence system to provide subjects with presence in remote environments through a mobile robot (Escolano et al., 2012a). That system relied on a P300-based BCI and a mobile robot with autonomous navigation and camera orientation capabilities. We further extended this system to the teleoperation of multiple robots (Escolano and Minguez, 2011) and performed initial tests in patients with amyotrophic lateral sclerosis (Escolano et al., 2010). 4Chapter 1. Introduction interest; (2) a novel method for the individualization of the brain patterns according to the combination of EEG recordings in two conditions (resting state and task-related activity); (3) a method for computing the baseline working level of the brain patterns per subject and session; and (4) a variety of methods and metrics to assess the effects of the NF on the individualized brain patterns targeted by the NF, the effects on non-targeted brain patterns (specificity of the procedure), the effects during the execution of cognitive tasks, and the effects at brain source level. In order to demonstrate the feasibility and experimental validity of this subject-specific NF approach, we carried out an implementation of a NF protocol of upper alpha up-regulation for cognitive enhancement, evaluated in three different NF studies involving healthy subjects, depressed patients and ADHD children. These studies investigated whether healthy, depressed and ADHD individuals could learn to increase the individual upper alpha power by means of NF, and whether these effects were related to behavioral effects on either cognition or clinical outcome. 1. The first study investigated the EEG and behavioral effects of a single NF session on healthy participants. Ten participants were assigned to the experimental group and nine to a sham-feedback control group. A NF protocol of upper alpha power enhancement over parieto-occipital locations was performed. This study showed increased upper alpha power in task-related activity (immediately after training), as well as increased pre-stimulus upper alpha power during the execution of a mental rotation task, apparent only for the experimental group. Both groups improved cognitive performance, with a more prominent improvement for the experimental group. However a single session seems to be insufficient to yield significant differences between groups. 2. The second study investigated the EEG and behavioral effects of eight NF sessions on depressed patients. 40 patients were assigned to the experimental group and 20 to a non-interventional control group. A NF protocol of upper alpha power enhancement over parieto-occipital locations was performed. This study showed increased upper alpha power in task-related activity (pre-post study) for the experimental group, not spatially or spectrally restricted to the trained parameter. A current density increase appeared at brain source level in alpha for the experimental group, localized in the subgenual anterior cingulate cortex (sgACC, BA 25). The experimental group showed increased performance as well as improved processing speed in a working memory test after the training, thus suggesting that the cognitive symptoms of depressed patients could be alleviated by this type of procedure. 5 3. The third study investigated the EEG and behavioral effects of 18 NF sessions on ADHD children. 20 children were assigned to the experimental group (preliminary uncontrolled study). A NF protocol of relative upper alpha power enhancement over fronto-central areas was performed. This study showed increased relative and absolute upper alpha power in task-related activity (pre-post study). While the pre-post study effects were mainly restricted to upper alpha, within-session analysis showed an absolute power decrease in slow-frequency oscillations (note that ADHD children commonly show an excess of slow-frequency activity). Parents rated a clinical improvement in children regarding inattention and hyperactivity/impulsivity, and neurophysiological tests showed an improvement in working memory. In summary, this thesis proposes the design of a subject-specific NF approach, which was subsequently implemented in a NF protocol of upper alpha up-regulation targeting cognitive enhancement in three different populations (healthy users, depressed patients and ADHD children). The results show that the technique was able to accommodate the large variability of the brain patterns among subjects and over sessions. In addition, these three populations were able to self-regulate the targeted brain patterns with a consequent improvement in cognitive performance and clinical outcome (in the case of ADHD children). Although the main contribution of the thesis is on the methods and on the implementation of the subject-specific NF approach, the NF studies herein presented are novel and the results extracted from them constitute an added contribution of this thesis. 6Chapter 1. Introduction 1.1 Structure and publications The contents of the thesis are organized as follows. •Chapter 2 provides some background information. We provide an overview of the EEG focusing on alpha and the individual alpha frequency, and the link between alpha and cognitive performance. We also provide an overview of NF containing its historical perspective, current protocols for cognitive enhancement, and the technical and methodological limitations of current NF techniques. •Chapter 3 presents the subject-specific NF technique developed in this thesis, which is currently protected by an European patent (Escolano et al., 2014a). •Chapter 4 investigated the NF effects in healthy subjects. This work was presented in Escolano et al. (2012b) and extended for a journal publication (Escolano et al., 2014d). •Chapter 5 investigated the NF effects in depressed patients. This work was presented in Escolano et al. (2013) and extended for a journal publication (Escolano et al., 2014b). •Chapter 6 investigated the NF effects in ADHD children. This work was published in Escolano et al. (2014c). •Finally, chapter 7 presents the conclusions of this thesis. 2|Background information This thesis develops a NF technique in terms of subject-specific signal processing methods and implements a NF protocol of upper alpha up-regulation for cognitive enhancement. Section 2.1 provides an overview of the EEG, focusing on alpha activity and the individualization of alpha frequency due to its well-known inter- and intra-subject variability, and we then describe the link between alpha and cognitive performance. Finally, Section 2.2 describes the principles of NF, provides a brief historical perspective, reports an overview of current NF protocols for cognitive enhancement, and reports the technical and methodological limitations of current NF techniques. 2.1 EEG The EEG reflects rhythmic fluctuations in the excitability of underlying neural populations (Niedermeyer and da Silva, 2005). EEG oscillations are commonly transformed to the frequency domain and divided into a set of frequency bands: delta, theta, alpha, beta and gamma (Figure 2.1). The activity in these frequency bands has been found to be related to brain states and functions (Niedermeyer and da Silva, 2005). As a brief summary, delta (0.5-4 Hz) mostly dominates deeper sleep states. Theta (4-7 Hz) displays characteristic patterns in states of drowsiness and sleep, and plays a maturational role in infancy and childhood (Niedermeyer and da Silva, 2005). Interestingly, an excess of theta activity in fronto-central sites is the most reliable EEG pattern in ADHD children (Barry et al., 2003). Alpha (8-12 Hz) is the dominant frequency in the human EEG and predominates during wakefulness relaxation (Klimesch, 1999). Beta (12-30 Hz) is associated with motor processing, showing a power decrease over sensorimotor areas during motor action (Pfurtscheller and Lopes da Silva, 1999). Low amplitude beta is associated with active cognitive engagement (Niedermeyer and da Silva, 2005). Faster oscillations as gamma (> 30 Hz) have not been as much studied as slower oscillations. These oscillations are modulated by cognitive processes 7 8Chapter 2. Background information such as attention and memory, and they have been suggested to be linked to binding and feature integration mechanisms (Senkowski et al., 2007). Time, 1 second Amplitude, 100 muV Beta Alpha Delta Theta Figure 2.1: Common EEG frequency bands: delta activity during deep sleep, theta in drowsy state, alpha in wakefulness relaxation state, and beta in wakefulness state. 2.1.1 Alpha activity Alpha rhythm is the dominant frequency in the human EEG, first recorded in the 1920s (Berger, 1929). This rhythm is characterized by a “peak” in the spectral analysis in the (8-12 Hz) frequency range, and predominates during wakefulness relaxation with closed eyes, best observable over posterior areas of the scalp (Klimesch, 1999). The parieto-occipital alpha rhythm is attenuated by eye opening, visual stimuli and by increased attentiveness (Palva and Palva, 2007). It also responds to motor tasks (Pfurtscheller and Lopes da Silva, 1999) and to different cognitive demands such as attention and memory tasks (Klimesch, 1999). Although there is not an universally accepted theory about the functional 2.2. Neurofeedback 15 participants performed a serial reaction time task with their non-dominant (left) hand. Participants who received NF immediately prior the task exhibited faster reaction times, suggesting a faster rate of learning. 2.2.3 Technical and methodological limitations of NF We next summarize some technical and methodological limitations of current NF techniques at both the online signal processing and the post-analysis of the effects on the EEG. Note that we do not intent to cover the extensive NF literature but rather to mention some of the major limitations. Calibration and online signal processing Some NF techniques record the EEG in only a reduced set of feedback locations (Nan et al., 2012, Alexeeva et al., 2012, Keizer et al., 2010). This restricts the potential signal processing methods to be subsequently applied to compute the feedback as well as the methods to be applied in the EEG post-analysis. Regarding the artifact filtering, some techniques do not have protection against artifacts or they implement a time-domain threshold for artifact detection, interrupting the feedback for some seconds when an artifact is detected (Zoefel et al., 2011). In the first case the accuracy of the feedback may be compromised and therefore hinder the learning of the selfregulation process, and in the second case the effective time of training may be reduced. In both cases it would be desirable filtering the artifacts rather than detecting them. Other techniques include either low frequency or high frequency oscillations (or both) as additional feedback for the subjects in order to try to reduce the blinking/eyes movement or muscular artifacts, respectively (Kober et al., 2014). This decision may make the task more complex for the subjects. A major concern of this thesis is the individualization of the signal processing methods to target subject-specific brain patterns. Traditional NF approaches use subject-generic filters to the decode the brain patterns of interest, commonly fixed frequency bands (Vernon, 2005). This approach may not to be able to deal with the EEG variability among subjects and sessions. Finally, some techniques set the baseline manually according to the trainer experience (Vernon, 2005), which may restrict the replicability of the results (training is dependent from the trainer). Post-analysis of the effects on the EEG NF literature lacks an extensive analysis of the electrophysiological effects, especially in non-trained EEG brain patterns, and during the execution of 16 Chapter 2. Background information cognitive tasks. As already mentioned, some NF techniques record the EEG in only a reduced set of feedback locations (Nan et al., 2012, Alexeeva et al., 2012, Keizer et al., 2010). This restricts the potentials methods to be applied and reduces the spatial resolution of the post-analysis. In fact many studies assess the effects on the trained brain patterns only. In addition, few metrics are commonly used to assess these effects. A recent review (Gruzelier, 2014) encouraged the use of several metrics such as the effects after the study (pre-post study effects), the immediate effects after the training trials (within-session effects) and the effects over the sessions (trend or learning curve over sessions). There are few examples of studies assessing the specificity of the NF effects on the EEG by measuring the effects on non-trained patterns, and those studies commonly restrict the analysis to a small number of pre-determined frequency bands (Zoefel et al., 2011, Nan et al., 2012). Furthermore, there are few studies assessing the effects of the NF on the EEG recorded during the execution of cognitive tasks. Some examples include the analysis of the EEG power time-course during the execution of a mental rotation task (Hanslmayr et al., 2005) and the analysis of event-related potentials during a short-term memory task (Kober et al., 2014). 3|The development of the neurofeedback technique This chapter describes the developed subject-specific NF technique. We first present the related problems and common concerns regarding NF techniques with an overview of the state of the art. We then present the design and implementation of a NF protocol of upper alpha up-regulation. Note that slight variations of this protocol (feedback locations and whether feedback was computed in absolute or relative power) were used in the three NF studies included in this thesis. On one hand, the NF protocol applied to healthy and depressed participants aimed to increase the absolute upper alpha power over parieto-occipital locations (chapters 4 and 5). On the other hand, the protocol applied to ADHD children targeted the relative upper alpha power over fronto-central locations (chapter 6). 3.1 Common concerns Artifact filtering. We refer as artifacts to the electrical signal recorded along with the EEG signal but that originates from non-cerebral sources. Biological artifacts such as eye blinks and muscular activity contaminate the EEG and can thus hinder the brain patterns of interest. This is the reason why artifact filtering is a common and important problem in EEG research (Delorme et al., 2007). The blinking artifact presents the stronger effects in low frequency oscillations (1-3 Hz) and may interfere higher frequencies as well, specially in frontal locations of the scalp, and is a natural body system with high occurrence rate. The muscular artifact induces strong (20-60 Hz) activity mainly at temporal locations (Delorme et al., 2007). In relation to NF literature, some techniques do not have protection against artifacts or they implement a time-domain threshold for artifact detection, interrupting the feedback for some seconds when an artifact is detected (Zoefel et al., 2011). Others include low and high frequency oscillations as additional feed- 17 18 Chapter 3. The development of the neurofeedback technique back so the subjects can potentially learn to reduce the blinking and muscular artifacts, respectively (Kober et al., 2014). In the first case the accuracy of the feedback might be compromised, in the second case the effective time of training might be reduced, and in the third case it might make the task more complex for the subjects and seems not well suited for some artifacts (e.g., eye blinking is a natural body system). Inter-subject variability of brain patterns. The EEG is a highly nonstationary signal that presents an inherent variability from subject to subject (Haegens et al., 2014, Neuper et al., 2005, Barry et al., 2003). As an example, Figure 3.1 displays the EEG power spectra of two subjects (depressed patient and ADHD child), illustrating the high variability in brain patterns, i.e., alpha frequency is around 10 Hz for the depressed patient and around 7.5 Hz for the ADHD child. This makes that most EEG systems need to adapt the technology to a particular subject as in BCIs, commonly executing a calibration step before the online operation (Pfurtscheller and Neuper, 2001). This individualization seems a “natural” approach to deal with the inter-subject variability. Most NF procedures do not target subject-specific brain patterns (Ros et al., 2014, Kober et al., 2013, Vernon, 2005), however there is a trend in alpha-based NF procedures in recent years towards the individualization of the alpha frequency (Zoefel et al., 2011, Hanslmayr et al., 2005, Nan et al., 2012, Alexeeva et al., 2012). Inter-session variability of brain patterns over NF training. The NF seeks for the self-regulation of some brain patterns of interest. The temporal evolution of these patterns (dynamics) over training sessions presents a high variability. As an example, Figure 3.1 displays the EEG power spectra of two subjects over NF sessions, illustrating the high inter-session variability, i.e., alpha power (for each subject) shows an increase over the course of the training program. This variability could be due to the non-stationary nature of the EEG (Niedermeyer and da Silva, 2005), to the physiological state of the subject and the task being performed (Klimesch, 1999), or could be a consequence of the self-regulation of these patterns. In this line, some studies have investigated the NF outcome depending on the mental strategy (Kober et al., 2013, Nan et al., 2012) and the confidence in the technology (Witte et al., 2013). Similarly, there is a growing interest in BCIs for neurophysiological predictors of performance (Blankertz et al., 2010). However, in both BCI and NF the factors influencing the outcome are unclear to date. The inter-session variability has been taken into account by many NF procedures. For example, traditional techniques adapted the baseline manually 3.1. Common concerns 19 Alpha variability: inter-subject & inter-session SESSION 02SESSION 01 resting state task-related SESSION 07 SESSION 15 SESSION 04 SESSION 08 Depressed patient (S17) ADHD child (S06) 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 Figure 3.1: EEG power spectra of two subjects (a depressive patient and an ADHD child) over three NF sessions (at the beginning, middle, and end of the study). Blue line depicts resting state activity whereas red line depicts task-related activity. Gray area denotes the traditional alpha (7-13 Hz) range. 20 Chapter 3. The development of the neurofeedback technique FP1 FP2 AFz F3 Fz F4 FCz C3 Cz C4 P3 Pz P4 O1 Oz O2 0 1 2 3 4 5 6 Figure 3.2: EEG signal (7 seconds) recorded from 16 electrodes. Raw signal is depicted in blue color, showing blinking artifacts, mostly apparent in anterior locations. ICA-filtered signal is superimposed in black color. according to the trainer experience (Demos, 2005). This may restrict the replicability of the results (experimenter dependent) and it makes difficult to implement a double-blind experimental design. Recent NF procedures perform an automatic inter-session calibration within each session (Zoefel et al., 2011, Kober et al., 2014, Nan et al., 2012, Alexeeva et al., 2012). 3.2 Design Artifact filtering. We designed an individualized real-time artifact filtering method using blind source separation (Hyvarinen, 1999). This method is well suited for blinking artifacts as they present a spatial pattern over the scalp. We re-computed this filter for each subject and session, this way the filter accommodates for not only the individual EEG patterns of each subject but also for other sources of variability such as slight differences in the sensors position. Figure 3.2 illustrates the result of this filtering. Filtering these artifacts can provide the subjects with more accurate feedback and more time of effective feedback (with regard to the detection strategy) in order to favor the learning of the self-regulation process. 3.2. Design 21 (A) Healthy subjects S02S01 resting state task-related S04 S09 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 S03 S12 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 S17S14 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 Figure 3.3: EEG power spectra over posterior locations of healthy subjects. Blue line depicts resting state activity whereas red line depicts task-related activity. Gray area denotes the traditional alpha (7-13 Hz) range. IAF was computed in alpha range, and it is shown as a vertical line, either in blue or red color, depending on whether it was computed in resting state or task-related activity, respectively. 22 Chapter 3. The development of the neurofeedback technique resting state task-related S04 S17 S10 S07 (B) Depressed patients S45S29 S27 S37 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 Figure 3.4: EEG power spectra over posterior locations of depressed patients. 3.2. Design 23 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 resting state task-related S02 S04 S05 S01 (C) ADHD children 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 S35S12 S06 S31 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 0 20 40 60 80 Figure 3.5: EEG power spectra over posterior locations of ADHD children. 24 Chapter 3. The development of the neurofeedback technique Inter-subject variability of alpha frequency. Alpha frequency has shown a high inter-user variability (Haegens et al., 2014, Klimesch, 1999), which may be even higher when dealing with some clinical populations such as ADHD (Barry et al., 2003, Lansbergen et al., 2011b). We designed a novel method for the individualization of the alpha frequency according to the combination of EEG recordings in two conditions (resting state and task-related activity). This method first computes the alpha frequency in task-related activity and, in those cases in which a clear alpha peak is not shown in the EEG power spectra, the alpha frequency is computed in resting state activity instead. This method may better detect the alpha frequency in comparison to strategies used in literature either using resting state or task-related activity only. Some procedures compute the alpha frequency in resting state conditions (Hanslmayr et al., 2005, Nan et al., 2012, Alexeeva et al., 2012), which may not take into account the inter-task variability of alpha frequency (Haegens et al., 2014). Other procedures compute it in task-related activity (Zoefel et al., 2011), which may throw misleading errors in some subjects due to the well-known alpha attenuation phenomenon during active brain processing (Klimesch, 1999). Figures 3.3, 3.4 and 3.5 illustrate this method. These figures show the EEG power spectra averaged across parieto-occipital locations, measured in resting state and task-related activity. EEG data from several participants of the studies are displayed, including healthy subjects, depressed patients and ADHD children. The first two rows (within each figure) show cases in which the alpha frequency could be accurately identified in task-related activity. Note that a subset of these subjects (e.g., healthy subjects 1, 2 and 12) display a high inter-task variability, showing faster alpha frequency in taskrelated activity than in resting state, which is in line with previous findings (Haegens et al., 2014). The last two rows show cases in which the alpha frequency was computed in resting state since no clear peak could be found in task-related activity. It is remarkable that some ADHD children showed a slow alpha peak (around 8 Hz) as found in other studies (Lansbergen et al., 2011b), suggesting that the individualization of brain patterns might be of special interest for this disorder. Inter-session variability of upper alpha power. Alpha oscillations have shown to respond to many different physical or mental tasks (Klimesch, 1999). We designed a method for computing the baseline working level of the NF per subject at the beginning of each session. This way the technique can take into account the inter-session variability of the EEG (dynamics). This is in line with recent NF procedures performing an automatic adjustment 4.2. Methods 31 performance and presents high levels of internal consistency and test-retest reliability (Tombaugh, 2006). The test scores were the number of errors and elapsed time. (ii) Rey Auditory Verbal Learning Test (RAVLT, Rey, 1964), Spanish version (Miranda and Valencia, 1997), evaluated retention and immediate evocation, verbal learning, remembering items after an interference task, and recognition (Lezak, 2004). The test score was the number of recognized words. (iii) Trail Making Test (TMT, Reitan, 1958) evaluated the information on visual search, scanning, processing speed, mental flexibility and executive functions. The test is composed of two parts: part A measured attention and concentration, and part B measured executive functions such as planning and mental flexibility. The scores were the elapsed time to complete each part of the test. (iv) Stroop Color-Word Test (STROOP, Stroop, 1992) evaluated attention, concentration, resistance to interference, and individual capacity to solve cognitive stress, inhibit interferences and process complex data (Lezak, 2004). The test score was the interference. The cognitive task was adapted from a visuospatial Spanish test (Yela, 1969) and EEG was recorded during its execution. In each trial, a target and a test figure were presented one above the other, and the subjects had to indicate (by pushing a button) whether the test figure corresponded to a rotated version of the target. Subjects were instructed to answer as quickly and accurately as possible. The test consisted of two phases of 25 trials each, with an inter-trial interval of 2.5 s. Each trial lasted 7.5 s and was composed of two time intervals: rest interval (-1.5, 0) s, in which a fixation cross was displayed on the center of the screen, and task interval (0, 6) s, in which the figures were presented for 6 s. Note that (t=0) s denotes figures onset. The test scores were the number of correct responses and reaction time. Responses within the task interval plus the inter-trial interval were taken into account and reaction times were computed for the correct responses. Two-way repeated-measures ANOVA was separately conducted for each score with the between-subject factor Group (NF, Control) and the withinsubject factor Time (Pre, Post). Post-hoc paired samples t-tests were performed for within-group (pre vs post) comparisons. EEG recording and neurofeedback procedure EEG data was recorded from 16 electrodes placed at FP1, FP2, F3, Fz, F4, C3, Cz, C4, P7, P3, Pz, P4, P8, O1, Oz and O2 (subset of the 10/10 system), with the ground and reference electrodes on FPz and on the left earlobe, respectively. EEG was amplified and digitized using a g.tec amplifier (Guger Technologies, Graz, Austria) at a sampling rate of 256 Hz, power-line notchfiltered at 50 Hz and (0.5-60) Hz band-pass filtered. EEG recording and the 32 Chapter 4. The effect of a single session of NF in healthy subjects NF procedure were developed using software of Bit&Brain Technologies, SL. EEG screenings EEG screenings were carried out immediately before and after the NF training on Day 2, as well as on Day 3 in order to asses immediate and one-day lasting effects on the EEG. For each EEG screening we recorded three-min of eyes closed resting state activity and three-min of eyes open task-related activity. In the latter, subjects faced a computer screen showing a square that changed saturation color randomly from gray to red or blue (gradually). Participants were instructed to count the number of saturation changes from gray to red as a cognitive challenge (Zoefel et al., 2011). Neurofeedback procedure The NF training focused on the increase of upper alpha (UA) power averaged over parieto-occipital locations (P3, Pz, P4, O1 and O2, referred to as feedback electrodes). The procedure consisted of two phases: calibration to individualize the training for each subject, and online training (5 trials of 5 min each). The description of the NF technique can be found in Chapter 3. Note that the calibration phase was computed on the pre-NF EEG screenings of Day 2 (resting state and task-related activity). Before the beginning of the study we recorded the EEG signal of a healthy subject performing the same NF procedure (i.e., pre- and post- EEG screenings and five training trials of the same duration), which was used as a “sham signal” for the control group. The software recorded the real EEG signal from the participants, but the signal processing methods were applied to the “sham signal” for the participants of the control group. Thus, all the participants in the control group received exactly the same feedback. Note that all the signal processing methods at both the calibration and online training steps were fully automatic and they did not require any supervision from the trainers. Thus trainers could be easily blinded to the group assignment, they simply had to enter an ID number in the software for each participant (ranging from one to the number of participants in order of arrival), which was previously assigned to a group by a technician who did not participate in the recordings. Trainers were shown the real EEG signal of participants in a computer screen with no additional information, and thus they could not guess the group assignment. 4.2. Methods 33 EEG analysis Offline EEG pre-processing The EEG data of each Day was cleaned from artifacts using a three-step procedure: filtering of the blinking component by FastICA (Hyvarinen, 1999), epoch rejection by a time-domain threshold (>150µV ) at any electrode, and epoch rejection by a frequency-domain threshold. In the latter step, we computed the power values for each epoch in the bands (1-4) Hz and (20- 30) Hz, commonly affected by ocular and muscular artifacts (Delorme et al., 2007), and outliers (z-score >2) at any electrode were removed. In the case of the EEG collected during the cognitive task, we applied an ICA filter, and we further applied epoch rejection in both time and frequency domains on a trial basis (instead of on an epoch basis). Analysis of EEG screenings and NF trials Immediate and one-day lasting effects were assessed in resting state and taskrelated activity. Immediate effects were measured as the power comparison between the prevs post- EEG screening of Day 2 (SCR.1 vs SCR.2, Figure 4.1). One-day lasting effects were measured as the power comparison between the pre-NF EEG screening vs EEG screening of Day 3 (SCR.1 vs SCR.3). In addition, training progress evaluated the power enhancement during training, measured as the power comparison between the baseline (task-related activity in pre-NF EEG screening) vs training trial five. We performed the analysis in the trained parameter (i.e., UA power) and an exploratory analysis in the following bands, based on Klimesch (1999): lower alpha 1, LA1= (IAF-4, IAF-2); lower alpha 2, LA2= (IAF-2, IAF); and lower beta, LB = (IAF+2, IAF+4). Analysis of event-locked EEG during the cognitive task The power in the pre- and postexecutions of the cognitive task was computed in rest interval (-1.5, 0) s, and task interval (0, 6) s. Note that (t=0) s represents the figures onset. Power was computed for each trial and averaged across trials. In addition, we computed the power desynchronization between rest and task intervals using two metrics: (i) absolute power desynchronization, computed as the power in task interval minus the power in rest interval, and (ii) event-related desynchronization (ERD), computed as the power in task interval minus the power in rest interval, normalized by the power in rest interval (Pfurtscheller and Lopes da Silva, 1999). Thus, the NF effects on the event-locked EEG were assessed by the prevs post- 34 Chapter 4. The effect of a single session of NF in healthy subjects power comparison in rest and task intervals, and by the prevs postpower desynchronization comparison measured using the absolute metric and the relative metric (ERD). We performed the analysis in the trained parameter and an exploratory analysis in LA1, LA2and LB bands. Statistical analysis Between-group statistical significance was assessed by independent samples t-tests on change scores. Paired samples t-tests were performed for withingroup (pre vs post) comparisons. Power values were log-transformed prior statistical testing. Regarding the exploratory analysis in LA1, LA2, UA and LB bands, a non-parametric randomization method using the t-max statistic was used to correct for the number of bands, i.e., to control the familywise type I error rate (FWER, Holmes et al., 1996). Following this method, the null distribution of the maximum absolute t-value across all bands was estimated by 5000 random permutations. Then the absolute observed t-value for each band was tested against the (1−α)th percentile of the null distribution. The FWER was set at α=.05. 4.3 EEG results This section analyzes the NF effects on the EEG by two sub-analyses: (i) immediate and one-day lasting effects in the EEG screenings and during training trials, and (ii) the event-locked EEG effects during the execution of the cognitive task. For each sub-analysis we first performed a prevs postcomparison of the effects on the trained parameter (UA power) and we then assessed the effects on surrounding bands (LA1, LA2, and LB) as an exploratory analysis. The results of the exploratory analysis were corrected for the multiple comparisons (number of bands) by a t-max randomization procedure (Holmes et al., 1996). Groups did not differ statistically in baseline IAF. Mean ±SD IAF in resting state was 9.8±0.2 Hz for the NF group and 10.3±0.3 Hz for the control group (independent samples t-test: t17 =−1.50, p =.15); and in task-related activity it was 10.2±0.2 Hz for the NF group and 10.6±0.3 Hz for the control group (t17 =−1.05, p =.3). Analysis of EEG screenings and NF trials We first assessed the NF effects in the trained parameter (UA power). Regarding the resting state, no significant difference appeared between groups 4.3. EEG results 35 SCR.1 TR.1 TR.2 TR.3 TR.4 TR.5 SCR.2 SCR.3 0.75 1 1.25 1.5 1.75 Norm.Power SCR.1 TR.1 TR.2 TR.3 TR.4 TR.5 SCR.2 SCR.3 0.75 1 1.25 1.5 1.75 Norm.Power (A) EEG SCREENINGS AND NF TRIALS: UA POWER NF group Control group NF PROCEDURE DAY task−rel. EEG scr. NF trials NEXT DAY NF PROCEDURE DAY NEXT DAY −5 0 5 µV2 Figure 4.2: NF effects on the EEG screenings and NF trials. (A) displays the UA power in the task-related EEG screenings (blue squares) and training trials (black dots). Values are normalized per subject to the power in the pre-NF EEG screening. Vertical bars indicate SEM. The gray line represents the training progress. Topoplots display the power difference (µV 2) with regard to the pre-NF EEG screening. 36 Chapter 4. The effect of a single session of NF in healthy subjects IMMEDIATE EFFECTS ONE-DAY LASTING EFFECTS NF group Control group TASK-RELATED ACTIVITY RESTING STATE TRAINING PROGRESS NF group Control group (B) EEG SCREENINGS AND NF TRIALS: POWER SPECTRA prespectra postspectra LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 LA1 LA2 UA LB 0 0.5 1 1.5 2 Figure 4.3: NF effects on the EEG screenings and NF trials. (B) displays the pre- and post- EEG power spectra for the immediate and one-day lasting effects on both resting state and task-related activity, as well as the training progress. Average and SEM power values are displayed (solid and dashed lines, respectively). Note that for illustration purposes, the pre- and postpower spectra were normalized per subject to the power in the pre-spectra IAF bin, and averaged across subjects. The (IAF-4, IAF+4) Hz frequency range is displayed (covering LA1, LA2, UA, and LB frequency bands). UA band is shaded in gray color. 4.3. EEG results 37 (on change scores) either immediately after the training or the following day. Figure 4.2 displays the UA power in task-related activity and during training. Task-related activity showed a between-group difference at statistical trend immediately after the training (t17 = 1.88, p =.077). Post-hoc t-tests showed a significant increase for the NF group (t9= 3.42, p < .01), with an average increase of 13.08%. These effects were not significantly sustained the following day. Training progress showed a between-group difference at statistical trend (t17 = 1.81, p =.089). Post-hoc t-tests showed a statistical trend for the NF group (t9= 1.94, p =.084), with an average increase of 51.22%. The higher effects in the latter metrics were found in posterior areas of the scalp. In addition to that, we measured the training progress as the trend (slope of a fitted regression line) of the power values of the baseline and five NF trials (see Figure 4.2). The average slope was 0.11 for the NF group (significantly higher than zero, t9= 1.86, p =.048) and 0.02 for the control group (non significant, n.s.). We assessed the EEG effects in three alpha sub-bands and lower beta (LA1, LA2, UA, LB). Figure 4.3 displays the pre- and post- EEG power spectra in immediate and one-day lasting effects, as well as the training progress. Note that the statistical results below reported were corrected for the multiple bands using a t-max randomization procedure (Holmes et al., 1996). No significant between-group differences (on change scores) appeared in resting state either immediately after training or the following day. Regarding the immediate effects in task-related activity, we found a significant between-group difference in LA2band (threshold t= 2.801,t= 3.69, p =.01). Post-hoc t-tests showed a significant increase for the NF group (threshold t = 2.98, t= 4.06, p < .01), with an average increase of 50.93%. No significant one-day lasting effects appeared in task-related activity. Finally, a betweengroup difference in training progress appeared in LA2band at statistical trend (threshold t= 2.74, t= 2.45, p =.085). Post-hoc t-tests showed a significant increase for the NF group (threshold t= 2.69, t= 4.14, p < .005), with an average increase of 74.55%. No significant prevs postchanges appeared for the control group. Analysis of event-locked EEG during the cognitive task We first assessed the NF effects in UA power. Figure 4.4 depicts the UA power time-course: absolute power time-course and ERD metric. Note that the absolute power time-course allows to observe the pre- and postpower 1The tthresholds reported throughout this study (for the t-max randomization procedure) were computed at (α= .05). 38 Chapter 4. The effect of a single session of NF in healthy subjects Time (s) µV2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 5 10 15 NF group Control group preexecution postexecution ERDAbs. power Time (s) ERD/S −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 −50 0 50 Time (s) ERD/S −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 −50 0 50 µV2 Time (s) µV2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 5 10 15 µV2 (A) EVENT-LOCKED EEG EFFECTS: UA POWER −5 0 5 µV2 ERD (%) ERD (%) REST TASK REST TASK Figure 4.4: NF effects on event-locked EEG during the pre- and postexecutions of a cognitive task. (A) displays the UA power time-course. Upper figures show the ERD measurement (Pfurtscheller and Lopes da Silva, 1999): UA power in each time instant was normalized to the power in rest interval. Bottom figures show the absolute UA power. Topoplots display the averaged pre- and postpower difference (µV 2) in both rest and task intervals. 4.3. EEG results 39 NF group Control group Initial scoresChange scores (B) EVENT-LOCKED EEG EFFECTS: ABS. POWER TIME-FREQUENCY MAPS −5 0 5 0 5 10 15 µV2 µV2 Figure 4.5: NF effects on event-locked EEG during the pre- and postexecutions of a cognitive task. (B) displays the absolute power time-frequency maps in the initial (preexecution) and change scores (prevs postexecution) for each group. The (IAF-4, IAF+4) Hz frequency range is displayed (covering LA1, LA2, UA, and LB frequency bands). 40 Chapter 4. The effect of a single session of NF in healthy subjects changes in each trial interval (rest, task). The UA power in rest interval showed a between-group difference (on change scores) at statistical trend (t17 = 1.96, p =.067). Post-hoc t-tests showed a significant increase for the NF group (t9= 3.97, p < .005), with an average increase of 16.61%. The higher effects were found in posterior areas of the scalp. No significant effects appeared for the task interval. A desynchronization pattern was apparent, showing an UA power decrease after the figures onset for both groups and pre- and postexecutions of the cognitive task (Figure 4.4). This pattern is in line with other studies performing similar mental rotation tasks (Hanslmayr et al., 2005, Klimesch et al., 2003). A between-group difference at statistical trend was found in absolute power desynchronization (t17 =−1.99, p =.063). Post-hoc t-tests showed a significant increase for the NF group (t9=−2.53, p=.032), with an average increase of 2.2µV 2. Note that a positive difference denotes an increase in desynchronization. We assessed the EEG effects in three alpha sub-bands and lower beta (LA1, LA2, UA, LB). Figure 4.5 displays the time-frequency analysis in the (IAF-4, IAF+4) Hz range, showing the initial and change (pre vs post) absolute power values for each group. Note that the statistical results below reported were corrected for the multiple bands (Holmes et al., 1996). We did not found significant differences between groups (on change scores) in any band and metric. We further conducted prevs postt-tests within each group. A significant increase appeared in LA2band for the NF group in rest interval (threshold t= 3.17, t= 4.52, p =.006), with an average increase of 29.9%; as well as in task interval (threshold t= 3.01, t= 4.10, p =.007), with an average increase of 12.1%. These effects in LA2power can be observed in Figure 4.5. No significant prevs postchanges appeared for the control group. 4.4 Behavioral results This section analyzes the NF effects on cognitive performance measured by a battery of psychological tests (targeting functions such as working memory, episodic memory, attention, concentration, and executive functions) and the cognitive task (mental rotation abilities). Table 4.1 summarizes the scores in the psychological tests and cognitive task. We assessed the between-group differences by the Group ×Time interaction in the ANOVAs. A significant effect appeared in part B of TMT test (F1,17 = 4.51, p=.049). Post-hoc t-tests showed that both groups improved performance (NF: t9=−4.26, p < .005; Control: t8=−3.52, p < .01), with a higher improvement for the NF group. No significant between-groups dif- 5|The cognitive effect of NF in major depression 5.1 Introduction Major depressive disorder (MDD) is a severe, chronic, mood disorder characterized by episodes of sadness, loss of interest and motivation, pessimism, and suicidal thoughts (DeRubeis et al., 2008). Depression affects each year 13 to 14 million citizens in the United States, with a lifetime prevalence of 16.2% (Kessler et al., 2003). MDD is estimated to become the second cause of burden of disease in 2030 (Mathers and Loncar, 2006). The current standard treatment for depression is antidepressant medication. Unfortunately with this treatment about 33% of patients fail to achieve remission (Anderson et al., 2008). Moreover, the patient compliance to the treatment can decrease due to the side effects such as sexual dysfunction, gastrointestinal problems, and weight gain (Brunoni et al., 2009). For this reason, the development of new treatments is constantly explored. In particular, recently there has been a renewed interest on neuromodulation techniques such as repetitive transcranial magnetic stimulation, transcranial direct current stimulation, and NF (DeRubeis et al., 2008, Brunoni et al., 2009). Most of the NF protocols applied to date to depressed patients are based on EEG findings of frontal asymmetry in the alpha frequency band, with depressed patients showing left hypoactivation (Henriques and Davidson, 1991, Davidson, 1998, 2004, Coan and Allen, 2004). These findings are interpreted as a dysfunction of the prefrontal cortex (PFC) and a predisposition towards negative emotions and behavioral withdrawal (Davidson, 2004, DeRubeis et al., 2008). While some studies support this theory (Gotlib, 1998, Lubar et al., 2003), controversial results have also been reported (Reid et al., 1998). Furthermore, some studies found asymmetry changes over recording sessions not related to the clinical symptoms of the patients (Allen et al., 2004a), thus its consideration as a marker for depression remains unclear to 47 48 Chapter 5. The cognitive effect of NF in major depression date (Allen and Cohen, 2010). Some NF studies have tried to reduce the alpha asymmetry in an attempt to alleviate the depressive symptoms, reporting promising results (Baehr et al., 1997, 2001, Hammond, 2000, 2005). However, to the best of our knowledge, none of the early studies were appropriately controlled. The first controlled study of this kind reported a reduction in depressive symptoms and an improvement in executive functions (Choi et al., 2011). The present work is based on a different rationale. Our objective is to alleviate the cognitive symptoms of depression. Cognitive deficits are core symptoms of depressive disorders with a clear-cut impact on social and occupational functioning, with patients showing decreased performance in working memory (WM) and attention, among others (Gotlib and Joormann, 2010, Castaneda et al., 2008, Austin et al., 2001). Furthermore, depressive patients show biases in the processing of emotional contents in WM (Gotlib and Joormann, 2010, Levens and Gotlib, 2010). For example, Levens and Gotlib (2010) performed an emotion n-back WM task in which depressed and control individuals had to match the valence of stimuli (happy, neutral or sad). Comparing the depressed vs control individuals they found in the 2-back task that depressed individuals were slower to match emotional stimuli in WM (regardless of the valence), and they integrated faster and removed slower the sad stimuli from WM. Recent evidences thus suggest that cognitive deficits may not only be correlates of depression, but they may also increase the risk for depression (Gotlib and Joormann, 2010, Levens and Gotlib, 2010). We hereby explore WM entrainment by means of a subject-specific NF protocol aimed to up-regulate the individual upper alpha power in parietooccipital locations. To the best of our knowledge, this is the first NF study exploring the cognitive effect of WM entrainment in patients with MDD. Eyes closed resting state and eyes open task-related activity were recorded before and after the training as well as pre- and postthe training trials within each session. A power EEG analysis and an alpha asymmetry analysis were conducted at the sensor level. Frequency domain standardized low resolution tomography (sLORETA) was used to assess the effect of training at the brain source level. Finally, a correlation analysis between the clinical/cognitive and EEG measurements was conducted at both the sensor and brain source level. The main cognitive variable was WM, measured along other variables such as attention, verbal memory and executive functions. 5.2. Material and Methods 49 5.2 Material and Methods Participants 74 participants diagnosed with major depressive disorder (MDD) were allocated to the NF group (n= 50) or to the control group (n= 24). Participants were not randomly allocated to groups (see discussion). Patients were recruited from different health centers in the city of Zaragoza (Spain). The inclusion criteria were age range (18-65 years), Spanish as the native language, diagnosis of MDD according to DSM-IV, and stable pharmacological/psychological treatment. Participants met DSM-IV criteria for MDD as assessed by their respective therapists. The exclusion criteria were diagnosis of comorbid disorders (e.g., schizophrenia, drug addiction, dementia). The experimental design was approved by the Ethical Review Board of the regional health authority and followed the Declaration of Helsinki. All participants signed an informed consent. Five participants dropped out the study due to the inability to perform the cognitive assessments or recording sessions, four of whom belonging to the NF group. Among the remaining participants who completed the study, six subjects in the NF group and three in the control group were excluded due to excessive artifacts in the EEG. Thus, the final sample consisted of 40 participants in the NF group and 20 in the control group. Demographic and clinical variables for both groups are reported in Table 5.1. There were no significant differences between groups in any variable. NF group Control group NF vs Control n= 40 n= 20 p-value Gender (male/female) 15/25 4/16 0.24 Age, years (mean ±SD) 53.70 ±10.87 49.50 ±10.18 0.16 Antidepressant medication (yes/no) 37/3 18/2 1.00 Comorbidity of anxiety (yes/no) 33/7 15/5 0.51 BDI-II score (mean ±SD) 23.70 ±13.51 22.25 ±11.74 0.68 PHQ-9 score (mean ±SD) 13.33 ±6.84 15.65 ±5.96 0.21 Table 5.1: Demographic and clinical variables of the participants at study entry. Between-group differences were assessed by Fisher’s exact test and independent samples t-tests for categorical and continuous variables, respectively. BDI-II = Beck Depression Inventory, Second Edition. PHQ-9 = Patient Health Questionnaire. 50 Chapter 5. The cognitive effect of NF in major depression 5 weeks NF GROUP CONTROL GROUP COGNITIVE ASM. EEG SCR. (FINAL) 60 min COGNITIVE ASM. EEG SCR. (INITIAL) 60 min NF TRAINING ( 8 sessions) ....... NF SES. 1 32 min NF SES. 8 32 min CLINICAL DIAGNOSTIC 60 min NF SESSION 6 min (2x3 min) EEG SCR. (PRE-NF) REST TASK 20 min (5 trials x 4 min/trial) NF TRIALS TR.1 TR. 2 ................ TR. 5 6 min (2x3 min) EEG SCR. (POST-NF) TASK REST Figure 5.1: Experimental design of the study. After an intake diagnostic both groups executed a pre- and postcognitive assessment and EEG screening within a five-week time interval. The NF group performed a total of eight NF sessions (two sessions per week), which were composed of a pre- and post- EEG screening (six min each) and five training trials (four min each). The EEG screenings included eyes closed resting state and eyes open task-related activity. Experimental Design The design of the study is shown in Figure 5.1. The severity of depressive symptoms was evaluated in a semi-structured interview using the Beck Depression Inventory (BDI-II, Beck et al., 1996) and the Patient Health Questionnaire (PHQ-9, Kroenke et al., 2001). After that, both groups performed a cognitive assessment and an EEG screening at the beginning and at the end of the study (five-week time interval). Cognitive assessments lasted approximately one hour. The NF group performed eight NF sessions over four weeks (two sessions per week). Each session was composed of five trials of four min each for a total of 20 min of training, and a pre- and post- EEG screening. For each EEG screening we recorded three-min of eyes closed resting state activity and three-min of eyes open task-related activity. In the latter, participants faced a computer screen showing a square that changed saturation color randomly from gray to red or blue (gradually). Participants were instructed to count the number of saturation changes from gray to red as a cognitive challenge (Zoefel et al., 2011). Cognitive Performance Pre- and postcognitive assessments were carried out at the beginning and at the end of the study. The main outcome variable was WM, which was measured using the Paced Auditory Serial Addition Task (PASAT, Gronwall, 1977) along with the processing speed. This test is sensitive to minimal changes in neurocognitive performance and presents high levels of internal consistency and test-retest reliability (Tombaugh, 2006). The test scores were the number of errors and elapsed time. Episodic memory, attention 5.2. Material and Methods 51 and executive functions were also assessed using the following tests: (i) Rey Auditory Verbal Learning Test (RAVLT, Rey, 1964) evaluated episodic memory and the test score was the number of recognized words. (ii) Stroop Color-Word Test (STROOP, Stroop, 1992) evaluated attention and concentration. The test score was the interference, which was standardized across age groups. (iii) Trail Making Test (TMT, Reitan, 1958) evaluated executive functions and was composed of parts A and B. The scores were the elapsed time to complete each part of the test. (iv) Fluency Verbal Test (FAS, Benton and Hamsher, 1976) evaluated verbal phonetic fluency. The test score was the number of evoked words. To determine statistical significance, two-way repeated-measures ANOVA (rm-ANOVA) was separately conducted for each score with the between-subject factor Group (NF, Control) and the within-subject factor Time (Pre, Post). Paired samples t-tests were performed for within-group (pre vs post) comparisons. EEG Recording and Neurofeedback Procedure EEG data was recorded from 16 electrodes placed at FP1, FP2, F3, Fz, F4, C3, Cz, C4, P7, P3, Pz, P4, P8, O1, Oz and O2 (subset of the 10/10 system), with the ground and reference electrodes on FPz and on the left earlobe, respectively. EEG was amplified and digitized using a g.tec amplifier (Guger Technologies, Graz, Austria) at a sampling rate of 256 Hz, power-line notchfiltered at 50 Hz and (0.5-60) Hz band-pass filtered. EEG recording and the NF procedure were developed using software of Bit&Brain Technologies, SL. The NF training focused on the increase of the individual upper alpha (UA) power averaged over parieto-occipital locations (P3, Pz, P4, O1 and O2, referred to as feedback electrodes). The description of the NF technique can be found in Chapter 3. Participants in the NF group were not given any mental strategy nor they were aware of the EEG trained parameter. Instead, they were encouraged to try different mental strategies guided by the feedback. To the best of our knowledge, the effect of different mental strategies in the ability to self-regulate brain activity is unclear to date, which is the focus of recent NF studies (Kober et al., 2013). Offline EEG pre-processing EEG data from the initial and final EEG screenings was carefully inspected for the presence of artifacts such as eye blinks, eye movements, body movements and EKG artifacts. Initially, the extended infomax ICA (Lee et al., 1999) was applied to the task-related activity to remove the eye blinking component. Then, both resting state and task-related activity were imported into 52 Chapter 5. The cognitive effect of NF in major depression EureKa! software (Congedo, 2002) to reject the contaminated data by visual inspection. Participants with at least 30 s of artifact-free data were included in the analysis1. EEG spectrum was computed following the same procedure as in the NF procedure. The EEG data of each NF session was cleaned from artifacts using a three-step automatic procedure: filtering of the blinking component by the extended infomax ICA (Lee et al., 1999), epoch rejection by a time-domain threshold (>200µV ) at any electrode, and epoch rejection by a frequency-domain threshold. In the latter step, we computed the power values for each epoch in the bands (1-3 Hz) and (20-30 Hz), commonly affected by ocular and muscular artifacts (Delorme et al., 2007). Then, we converted the log-transformed power values to z-scores and removed the outliers (>2.5) at any electrode. Note that the automatic procedure was only used to compute the alpha power in parieto-occipital locations, and no significant differences were found between the two procedures in either resting state (paired samples t-test: t59 =−0.77, p =.44) or task-related activity (t59 = 0.26, p =.79). EEG power was then computed through a short-term fast Fourier transform (FFT) with 1 s hamming window, 30 ms of overlapping, and zero-padded to 1024 points (0.25 Hz resolution). Power EEG Analysis in the trained parameter Power EEG analysis was conducted in the trained parameter: power in the individual UA band, averaged across the feedback electrodes (P3, Pz, P4, O1, O2). The pre-post enhancement was measured as the power change between the initial and final EEG screenings in resting state and task-related activity for both groups. The across- and within-session enhancement were also measured (for the NF group). Across-session enhancement was assessed by a linear trend analysis of the power values in the pre-NF screenings of all sessions and the final screening. The within-session enhancement comprised two measurements, computed in the power values averaged across the NF sessions: a power change comparison between the pre- and post- EEG screenings, and a linear trend analysis of the power values in the pre-NF taskrelated EEG screening and the five training trials. To determine statistical significance of pre- and postcomparisons, log-transformed power values were entered into a two-way rm-ANOVA with the factors Group (NF, Control) and Time (Pre, Post). Paired samples t-tests were performed for within-group (pre vs post) comparisons. Trend analysis consisted in the computation of the slope of a fitted regression line for each participant, and a t-test to test 1Some studies found reliable test-retest correlation coefficients in FFT frequency analysis performed on segments of 60, 40, or 20 s total length (Salinsky et al., 1991). 5.2. Material and Methods 53 the hypothesis of a null slope. The type I error was set at α=.05. Power EEG Analysis in the Sensor x Frequency domain Power analysis was conducted for all sensors and frequencies in the (1- 30 Hz) interval, separately applied for the resting state and task-related activity. The NF training effects were measured as the log-transformed power spectra comparison between the initial and final EEG screening: a between-group comparison (NF vs control group) on the change power values, and a within-group comparison (pre vs post) for each group. A cluster-based non-parametric randomization method (Nichols and Holmes, 2002, Maris and Oostenveld, 2007) was used, as implemented in the Fieldtrip toolbox (FC Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands; see http://www.ru.nl/fcdonders/fieldtrip). This method first computes the difference between two conditions by performing t-tests in the (sensor, frequency)-pairs. Those pairs exceeding a threshold qare clustered on the basis of spatial and spectral adjacency, and then cluster-level statistics are calculated as the sum of the t-values within every cluster. The threshold was set to (q=.05) for resting state and to (q=.01) for task-related activity. Finally, the significance probability at the cluster-level was estimated by a permutation method (Pesarin, 2001). The distribution of the cluster values was constructed under the null hypothesis by 5000 random permutations and then the observed values were tested against the (1 −α)th percentile of the null distribution. This method controls for the type I error rate and corrects for multiple comparisons both across sensors and frequencies. The type I error at cluster-level was set to α=.05. Alpha Asymmetry Analysis Initial and final scores of alpha (8-12 Hz) asymmetry were computed in resting state and task-related activity. EEG data was re-referenced to Cz and asymmetry scores were computed as the normalized power difference between homologous right- and left-side locations, (R-L)/(R+L). See Allen et al. (2004b) for a review on methodological considerations. This score indicates the relative activation of the left over right locations. Thus positive scores indicate left-lateralized activation, i.e., more power over right-side locations due to the inverse relation between alpha power and brain activation (Coan and Allen, 2004). Alpha asymmetry scores were computed in five areas of the scalp: prefrontal (FP: FP2-FP1), frontal (F: F4-F3), central (C: C4-C3), parietal (P: (P7+P3)/2-(P8+P4)/2) and occipital (O: O2-O1). Independent samples t-tests were conducted to test for between-group dif- 54 Chapter 5. The cognitive effect of NF in major depression ferences in initial scores. We applied t-tests on the initial scores (for each group and area of the scalp) to test for null asymmetry scores. Two-way rm-ANOVA with the factors Group (NF, Control) and Time (Pre, Post) was separately conducted for each area of the scalp to test for pre-post study changes. Bonferroni correction was applied to correct for multiple areas so as to keep the FWER at α=.05. EEG Analysis at the Brain Source Level Frequency domain standardized low resolution tomography, sLORETA (Pascual-Marqui, 2002, 2007) was used to estimate the current density of brain sources in resting state and task-related activity. The current density changes were compared between groups (NF vs control group) as well as within groups (pre vs post) for each group. EEG data was re-referenced to a common average reference and Fourier cross-spectral matrices were computed for the following frequency bands: delta (1-4 Hz), theta (4.5-7 Hz), alpha (8-12 Hz), beta1 (12-15 Hz), beta2 (15-20 Hz) and beta3 (20-30 Hz). These bands were defined according to the results obtained in the clustering analysis at the sensor level (section 5.2). After that, sLORETA estimated the current density values in 6239 voxels (5 mm3spatial resolution). sLORETA applies the boundary element method on the MNI-152 (Montreal Neurological Institute, Canada) template of Mazziotta et al. (2001), and the anatomical labelling is based on probabilities returned by the Daemon Atlas (Lancaster et al., 2000). Finally, current density values were log-transformed and statistical significance of each voxel was determined by a non-parametric randomization procedure using the t-max statistic to control the familywise type I error rate (FWER, Holmes et al., 1996). Following this procedure, the null distribution was estimated by 5000 random permutations under the null hypothesis of the maximum absolute t-value across all voxels. Then the absolute observed t-value for each voxel was tested against the (1−α)th percentile of the null distribution. Bonferroni correction was applied to correct for multiple bands so as to keep the FWER at α=.05. Correlation Analysis: EEG vs behavioral variables Spearman correlation was employed to test for a correlation in the initial scores between the clinical/cognitive and EEG variables, as well as in the change scores between the cognitive and EEG variables (clinical variables were not measured after NF training). A total of two clinical variables and seven cognitive variables were tested (section 5.2). The EEG variables were assessed in resting state and task-related activity and can be divided into 5.3. Results 55 Test NF CONTROL Paired-samples t-test ANOVA Scores Pre Change Pre Change NF CONTROL G×T mean (SEM) mean (SEM) mean (SEM) mean (SEM) p-value p-value p-value PASAT # errors 13.45(1.09) −3.32(0.64) 11.59(2.22) −0.41(1.23) <.001 .742 .024 time (s) 253.23(13.25) −39.55(8.05) 249.76(20.40) −3.71(16.08) <.001 .821 .031 RAVLT # recognized words 12.25(0.43) 0.90(0.27) 11.55(0.65) 0.35(0.65) .002 .597 .363 STROOP interference 51.64(1.23) 1.74(1.20) 54.60(1.80) 0.20(1.52) .154 .897 .443 TMT part A, time (s) 44.66(3.03) −6.37(3.26) 41.35(3.02) −3.40(2.43) .059 .178 .543 part B, time (s) 101.50(14.57) −16.63(5.83) 84.90(11.60) −11.50(10.55) .007 .289 .645 FAS # evoked words 43.74(1.87) 3.10(1.19) 38.70(2.68) 1.10(1.54) .013 .482 .319 Table 5.2: Cognitive performance at the beginning and at the end of the study for each group. Mean ±SEM scores in precognitive assessments are shown, as well as the change scores (post-pre). The p-values of the paired samples t-tests are shown for each group (pre vs post changes), as well as the p-values of the Group × Time interaction in the rm-ANOVAs. Significant effects are marked bold (p < .05), statistical trends are underlined (p < .1). PASAT = Paced Auditory Serial Addition Task, RAVLT = Rey Auditory Verbal Learning Test, TMT = Trail Making Test, FAS = Fluency Verbal Test. two groups: power variables at the sensor level, and current density variables at the brain source level. In both cases, the analysis was conducted in the aforementioned frequency bands (delta, theta, alpha, beta1, beta2, beta3). In the case of the sensor level, power values in each band were averaged across five areas: prefrontal (FP: FP1, FP2), frontal (F: F3, Fz, F4), central (C: C3, Cz, C4), parietal (P: P7, P3, Pz, P4, P8) and occipital (O: O1, Oz, O2). In the case of the brain source level, a randomization procedure was used to control the FWER (Holmes et al., 1996). 5000 random permutations were performed to construct a distribution of the maximum of the absolute rvalue across all voxels under the null hypothesis. In both cases the Bonferroni correction was applied to keep the FWER at α=.05. 5.3 Results Cognitive Performance Table 5.2 summarizes the scores in the cognitive assessments. A significant Group ×Time ANOVA interaction appeared in the PASAT test (# errors: 56 Chapter 5. The cognitive effect of NF in major depression F1,46 = 5.42, p =.024; time: F1,46 = 4.97, p =.031), showing an improvement in WM performance and processing speed for the NF group only (# errors: t30 =−5.21, p < .001; time: t30 =−4.91, p < .001). Cohen’s deffect size (Cohen, 1988) revealed a medium-large effect for both the number of errors (d=.703) and elapsed time (d=.673). Note that 9 participants in the NF group did not complete the PASAT test in the initial assessment due to excessive cognitive effort, 4 of whom completed the final assessment. 3 participants in the control group did not complete the PASAT test in both initial and final assessments. No significant ANOVA interaction appeared in the other variables. Regarding the within-group (pre vs post) changes, the number of recognized words increased in the RAVLT test for the NF group (t39 = 3.28, p < .005). Interference score in the STROOP test was not significantly modified for any group. Both parts of the TMT test improved for the NF group (part A: t37 =−1.95, p =.059; part B: t37 =−2.85, p < .01). The NF group also improved the number of evoked words in the FAS test (t38 = 2.61, p < .05). No significant changes were found for the control group. Power EEG Analysis in the trained parameter The analysis of the pre-post enhancement in the trained parameter (power in the individual UA band averaged across the feedback electrodes: P3, Pz, P4, O1, O2) revealed a no significant Group ×T ime ANOVA interaction for the resting state activity. However, a statistical trend between the pre and post measurements was found in the NF group (t39 =−1.72, p =.093), with an average increase of 22%. Regarding the task-related activity, a significant Group ×Time interaction appeared (F1,58 = 14.88; p < .001). Post-hoc t-tests showed a significant pre vs post difference for the NF group only (t39 =−5.44, p < .001), with an average increase of 56%. No significant change was found for the control group in either resting state or task-related activity. Note that groups did not differ statistically in initial IAF. Mean ±SD IAF measured in resting state activity was 9.81 ±0.17 Hz for the NF group and 9.79 ±0.25 Hz for the control group (t-test for independent samples, t58 =.075, p=.94); in task-related activity it was 9.71±0.19 Hz for the NF group and 9.64 ±0.24 Hz for the control group (t58 =.21, p=.83). Furthermore, IAF did not change significantly after the NF training for either group. The across- and withinsession enhancement was measured for the NF group (see Figure 5.2). Trend analysis revealed a significant UA power increase across the NF sessions in both resting state (t39 = 2.56, p =.014) and task-related activity (t39 = 4.04, p < .001). Regarding the within-session enhancement, a significant power increase between the pre- and post- NF screenings appeared in resting state (t39 =−3.10, p < .005), with an average 5.4. Discussion 63 Resting state act. FP F C P O −0.1 −0.05 0 0.05 0.1 + - + - FP F C P O −0.1 −0.05 0 0.05 0.1 * + - Task-related act. NF Control Figure 5.6: Alpha asymmetry scores measured in prefrontal (FP: FP2-FP1), frontal (F: F4-F3), central (C: C4-C3), parietal (P: (P7+P3)/2-(P8+P4)/2) and occipital (O: O2-O1) areas of the scalp. Positive asymmetry scores denote left over right activation, i.e., more alpha power in right hemisphere locations. ∗significant effect (p < .05), ±statistical trend (p < .1). Note that statistical results are not corrected for the multiple areas. ity. Since stronger effects were found in task-related activity in comparison to resting state, we adapted the qthreshold to each recording condition to get clearer sensor x frequency maps. Please note that qthreshold does not determine the type I error at cluster-level, which was set to (α=.05) for both conditions. Significant clusters were found for the NF group only, showing a power increase after the NF training. The strongest effect in resting state activity appeared in the lower part of alpha (8-10 Hz). Task-related activity showed stronger effects in the (4.5-20 Hz) range, covering theta, alpha and the lower part of beta. These effects were apparent in anterior, central and posterior locations. Our results show the common finding that the effects of the NF training on the spectral power are not spatially or spectrally restricted to the trained locations and frequency bands, instead the training engenders profound changes of the homeostatic properties of the brain involving several locations and frequencies (Hughes and John, 1999). The strong effect in task-related activity illustrates the importance of recording EEG in several conditions to provide additional information of the underlying brain processes. This is in contrast to the common practice to study only the resting state, either eyes closed or eyes open. Alpha asymmetry scores did not differ between groups at study entry and it did not change pre-post the study for any group. While previous studies have found a frontal asymmetry in depression (Davidson, 2004) the current study can-not assess this finding since we would need to compare our results to a control group of non-depressed participants. We investigated the 64 Chapter 5. The cognitive effect of NF in major depression initial alpha asymmetry scores for null scores, measured in five areas of the scalp (prefrontal, frontal, central, parietal and occipital). Strong asymmetry scores were found although they failed to reach statistical significance after strict control of the type I error (Bonferroni correction). Here we summarize the significant results when not correcting for the multiple areas (see Figure 5.6). Significant scores were found in posterior areas (parietal and occipital). Left-lateralized activation (more alpha power over right locations) was found in resting state in the parietal area at a trend level for both the NF group (t39 = 1.76, p =.086) and the control group (t19 = 1.86, p =.077). These results are in line with previous studies (Kemp et al., 2010). Interestingly, the opposite effect was found in task-related activity. Right-lateralized activation (more alpha power over left locations) was found in task-related activity in the occipital area at a trend level for the NF group (t39 =−1.98, p =.054), significant for the control group (t19 =−2.71, p =.014). Please note that EEG data was recorded using a single earlobe reference, but re-referenced offline to Cz. The most common approach to measure brain asymmetry to date is the use of computer-averaged ears/mastoids reference, or Cz reference (Davidson, 1998, Coan and Allen, 2004, Allen and Cohen, 2010). Although there is no agreement about a preferred montage, reference placement might be a critical issue (Hagemann et al., 2001, Allen et al., 2004a, Davidson, 2004), with unilateral references being discouraged. Thus, the alpha asymmetry effects herein reported should be taken with caution and confirmed in future studies using one of the aforementioned recording approaches. The analysis at the brain source level (using sLORETA) revealed significant changes in current density after the NF training for the NF group only. The stronger effect was found in task-related activity for the alpha band (8- 12 Hz), localized in the subgenual ACC (sgACC, BA 25) extending to the entorhinal cortex (BAs 28, 34), ventromedial prefrontal cortex (BA 32) and orbitofrontal cortex (BA 11). Notice that we are controlling for the familywise type I error rate using a conservative method for correcting also for the multiple frequency bands (t-max permutation tests were further corrected using a Bonferroni correction for the multiple bands). Strong effects were also found in other brain structures although they failed to reach statistical significance after such a strict control of the type I error. Here we summarize the significant results when not correcting for the multiple frequency bands, which appeared for the NF group only. Regarding the resting state activity, a current density increase appeared in the pregenual ACC, pgACC (BA 24; XYZ = −5,35,10; t= 3.25) measured in alpha band. Regarding the taskrelated activity, a current density increase appeared in the sgACC (BA 25; XYZ = 0,5,−5; t= 3.03) measured in delta (1-4 Hz), in the pgACC (BA 5.4. Discussion 65 24; XYZ = −5,30,0; t= 3.01) in theta (4.5-7 Hz), and in the pgACC (BA 24; XYZ = −5,25,15; t= 3.27) in beta1 (12-15 Hz). Correlation Analysis: EEG vs behavioral variables A strong correlation was found between the processing speed (elapsed time variable in the PASAT test) and the power in task-related activity measured in the beta frequency band. The power EEG analysis revealed that higher power in beta2 band (15-20 Hz) in parietal locations positively correlated with slower processing speed at study entry. We also found that the power increase in beta band (specifically in each one of the three beta sub-bands analyzed, ranging from 12 to 30 Hz) in prefrontal locations positively correlated with the improvement in processing speed (pre vs post NF training). These results suggest that beta band is related to processing speed in different ways according to the area of the scalp involved. Furthermore, a strong correlation was found at the brain activity recorded in different scalp locations: the increase in current density measured in beta3 band (20-30 Hz) positively correlated with the improvement in processing speed (pre vs post NF training). This correlation was localized in the pgACC (BA 24) and sgACC (BA 25) extending to the ventromedial prefrontal cortex (BA 32) and further into the orbitofrontal cortex (BA 11) and frontopolar cortex (BA 10). Beta band activity has been traditionally linked to somatosensory and motor functions but its functional role is not well understood to date (see Engel and Fries, 2010, for a review). Regarding its perceptual-cognitive role, it has been recently associated with the active continuation of the current cognitive set in tasks involving a strong top-down component (Engel and Fries, 2010). According to this theory, the brain makes predictions in beta oscillations about what it will encounter in the internal and external environment, and updates this by measuring prediction errors, which are encoded by gamma oscillations (Arnal and Giraud, 2012). A large body of research has highlighted the relation of the ACC in error monitoring and prediction of errors (Holroyd et al., 2009, Rushworth and Behrens, 2008). The improvement in processing speed after NF training may be thus explained by this relation. The involvement of the Anterior Cingulate Cortex The ACC is known to be linked to cognitive and emotional processes (Bush et al., 2000). Based on cytoarchitecture, lesion and electrophysiological studies the ACC has been divided into two major functional sub-divisions: a cognitive and an affective sub-divisions (Bush et al., 2000). The cognitive 66 Chapter 5. The cognitive effect of NF in major depression division is localized in the dorsal ACC (areas 24’, 32’), and the affective one is localized in the rostral-ventral ACC (rostral areas 24 and 32, and ventral areas 25 and 33). Functional differences in the ACC between depressed and non-depressed subjects have been repeatedly reported (Davidson et al., 2002, Mayberg, 1997, Pizzagalli et al., 2001). On one hand, decreased activation has been reported in the dorsal ACC (dorsal region of area 32; areas 24’, 32’). On the other hand, increased pre-treatment activation has been found in responders (vs non-responders) to antidepressant medication in the rostral and ventral ACC (including pregenual areas 24 and 32), and it has been suggested as a predictor of treatment response (Pizzagalli et al., 2001, Mayberg et al., 1999). Pizzagalli et al. (2001) compared depressed participants showing better response to nortriptyline treatment to responders showing worse response. Participants showing a better response had higher pre-treatment theta (6.5- 8 Hz) activity localized in the rostral ACC (BA 24, 32), estimated using sLORETA. In this line, the current study showed a current density increase in the rostral ACC (BA 24; XYZ = −5,30,0; t= 3.01) measured in the theta band (4.5-7 Hz), suggesting that the present NF protocol may increase the response to antidepressant medication. Regarding the correlation between the cognitive and EEG variables, the pregenual ACC (BA 24) and subgenual ACC (BA 25) were positively correlated with the improvement in processing speed. Interestingly, the subgenual ACC has not been traditionally linked to cognitive processing. However, recent studies demonstrate that there is an overlap of autonomic, sensorimotor, affective and cognitive processing in the midcingulate gyrus (Beissner et al., 2013) suggesting that the functionalanatomical parcellation of the cingulate gyrus is not as simple as has been assumed. The present study suggests that the subgenual ACC, apart from its well-known involvement in autonomic (Beissner et al., 2013) and emotional (Bush et al., 2000, Mayberg et al., 1999) regulation, it also is implicated in cognitive processing. Thus analogous to the midcingulate gyrus also the subgenual ACC may be involved in multiple overlapping functions. Limitations Due to the novelty and the exploratory character of the study, the control group designed in the present study was not optimal. On one hand, the number of subjects in the NF and control group was not balanced. We decided to use an allocation ratio 2:1 between the NF and control group since we were mainly interested on evaluating the effects of the NF on the EEG. On the other hand, participants were not randomly assigned to the experimental condition nor they were blinded with respect to the experimental condition. No randomization was used because the study was conducted in three dif- 5.5. Conclusions 67 ferent phases according to the participants and researchers availability (two phases for the experimental group, one phase for the control group, with each phase involving around 25 participants). Regarding the demographic characteristics, around 90% of the participants of the study followed a stable pharmacological antidepressant treatment during the study, 75% of whom consisted either on selective serotonin reuptake inhibitors (SSRI), benzodiazepines or serotonin-norepinephrine reuptake inhibitors (SNRI). Also, 80% of the participants presented comorbidity with anxiety. Further research is needed to elucidate to what extent the obtained results can be translated to a drug-free population or to a population without anxiety symptoms. The averaged severity of depression was “moderate” for both the NF and control group according to the BDI-II. However, the averaged PHQ-9 scores were slightly lower for the NF group (“moderate”) than for the control group (“moderately severe”). Nonetheless, there were no significant differences between groups in severity of depression as measured by either BDI-II or PHQ-9. The current study estimated the effects at brain source level using sLORETA (MNI-152 template) with only 16 electrodes. The spatial resolution and precision of sLORETA method could be improved by computing individual head models based on magnetic resonance imaging (MRI). A comparison of the cognitive performance with healthy subjects would have been an interesting analysis to estimate the significance of the observed effects. This analysis should be taken into account for future studies. Finally, although we have stated that the present NF protocol has the potential to improve depressive symptoms, such hypothesis can-not be corroborated in the present study since clinical variables were not assessed after NF training. While the control group used in the present study can account for practice effects in the cognitive assessments and for electrophysiological changes before and after the study, the motivation and expectations might be higher for the experimental group. Thus, the placebo effect can-not be ruled out with this experimental design. Due to the positive cognitive effects, the present NF protocol should be evaluated in future studies using stricter control conditions such as active control conditions (e.g., psychotherapy) or sham feedback. However, a sham feedback control condition may lead to ethical concerns when effective standard treatments are available (La Vaque and Rossiter, 2001). 5.5 Conclusions This study showed that depressed patients can enhance upper alpha activity through NF training in comparison to a non-interventional control group. These effects in the power spectra were not spatially or spectrally restricted 68 Chapter 5. The cognitive effect of NF in major depression to the trained parameter, instead stronger effects appeared in the (4.5-20 Hz) range for task-related activity, covering anterior, central and posterior scalp locations. The experimental group showed a current density increase in task-related activity for the alpha band (8-12 Hz), localized in the subgenual ACC (sgACC, BA 25). A positive correlation was found for the experimental group between the improvement in processing speed and the increase of beta power at both the sensor and brain source level. Finally, the experimental group showed improved WM performance as well as processing speed, thus suggesting that the cognitive symptoms of depressed patients could be alleviated by this type of procedure. Up to now, we have investigated the effects of the subject-specific NF protocol (upper alpha up-regulation) over parieto-occipital locations in healthy and depressed participants. Next study will investigate the effects of relative upper alpha up-regulation over fronto-central location in ADHD children. This study will extend the current analysis of the specificity by considering the subject-specific frequency bands (i.e., analysis centered to the IAF) and by assessing its within-session effects. 6|The effect of NF in ADHD children 6.1 Introduction Attention-deficit/hyperactivity disorder (ADHD) is a behavioral disorder characterized by symptoms of inattention, impulsivity and hyperactivity according to DSM-IV (American Psychiatric Association, 1994). This disorder is one of the most common psychiatric disorders of childhood, affecting up to 5% of children worldwide (Polanczyk et al., 2007), presenting about 40-60% persistence in adolescence and adulthood (Faraone et al., 2006). Deficits in executive functioning, working memory and response inhibition have been repeatedly reported (Barkley, 1997, Martinussen et al., 2005, Castellanos and Tannock, 2002). The most accepted treatments for ADHD are stimulant medication and behavior therapy (Barkley, 1997). Stimulant medication has emerged as the primary treatment for the core symptoms of ADHD, however some children do not respond to medication or suffer from side effects such as headache, dizziness, insomnia, anxiety and gastroenterological problems (Graham et al., 2011). In addition to that, the long-lasting effects of both stimulant medication and behavior therapy are uncertain, with some studies reporting limited effects (Wang et al., 2013, Molina et al., 2009). NF is a promising alternative treatment for ADHD (Arns et al., 2014, Loo and Makeig, 2012). The rationale behind NF training in ADHD is the electrophysiological evidence collected over the last decades of abnormal brain oscillations in comparison to normal controls (see Barry et al. (2003), Loo and Makeig (2012) for reviews). The most reliable EEG pattern in ADHD to date is an excess of theta activity (4-8 Hz) in fronto-central sites, measured in resting state (Barry et al., 2003, Snyder and Hall, 2006, Clarke et al., 2001). Reduced alpha (8-13 Hz) and beta activity (13-30 Hz) have been commonly reported as 69 70 Chapter 6. The effect of NF in ADHD children well, thus theta/beta and theta/alpha ratios have been traditionally considered reliable measures to discriminate between ADHD and normal individuals (Barry et al., 2003, Snyder and Hall, 2006). In this direction, NF studies have mostly used standardized protocols to “correct” the aforementioned abnormal EEG patterns. The most used protocol is theta suppression/beta enhancement, usually enhancing SMR simultaneously (Loo and Makeig, 2012, Monastra et al., 2006, Arns et al., 2009). An extensive evaluation of standardized NF protocols has been performed during the last 40 years in ADHD children, with recent reviews pointing out their effectiveness (Arns et al., 2014, Loo and Makeig, 2012, Heinrich et al., 2007). Despite the positive results, these protocols may not be able to account for the large EEG heterogeneity in ADHD (Loo and Makeig, 2012, Arns et al., 2008). In addition, recent findings challenge the theta/beta ratio as a marker for ADHD, which was found increased in only 20-30% of ADHD individuals (Arns et al., 2013, 2012). This may be partially due to a subgroup of 10-15% ADHD individuals showing increased (rather than decreased) beta activity (Clarke et al., 2013, 2001). Thus, individualized approaches may better cope with the EEG heterogeneity and improve the clinical outcome (Arns et al., 2014), which is the direction followed by some recent NF studies (Arns et al., 2012, Lansbergen et al., 2011a, Logemann et al., 2010). The current study evaluates a subject-specific NF protocol for ADHD children. This NF protocol aims at enhancing the relative upper alpha power in fronto-central sites, individually determined for each child using the individual alpha frequency (IAF, Klimesch, 1999) as an anchor point. On one hand, this protocol has the potential to decrease the excess of absolute theta power (most reliable EEG pattern in ADHD to date) and the excess of slow frequency oscillations in general. On the other hand, this protocol builds upon the positive results of alpha-based protocols in cognitive enhancement, mainly evaluated in healthy users (Gruzelier, 2013). Thus, this NF protocol is meant to target the cognitive deficits of ADHD individuals. This study reports an open-label pilot study with 20 ADHD children who underwent 18 NF sessions. EEG was recorded pre- and posttraining, and pre- and postthe training trials within each session, in both eyes closed resting state and eyes open task-related activity. A power EEG analysis assessed prepost study and within-session effects, in the trained parameter and in all the sensors in the (1-30) Hz spectral range. Learning curves over sessions were assessed as well. Clinical and neurophysiological variables were measured pre- and posttraining. 6.2. Methods 71 2 months BEHAVIORAL ASM. EEG SCREENING 60 min BEHAVIORAL ASM. EEG SCREENING 60 min NF TRAINING (1 8 sessions) ....... NF SES. 1 32 min NF SES. 18 32 min INTAKE EVALUATION 60 min NF SESSION 6 min (2x3 min) EEG SCR. (PRE-NF) REST TASK 20 min (5 trials x 4 min/trial) NF TRIALS TR.1 TR. 2 ................ TR. 5 6 min (2x3 min) EEG SCR. (POST-NF) TASK REST Figure 6.1: Experimental design of the study. After an intake evaluation, the children carried out an initial and final behavioral assessment (clinical and neuropsychological tests) and EEG screening within a two-months time interval. The NF training consisted of 18 sessions, which were composed of five training trials (four min each) and a pre- and post- EEG screening. The EEG screenings included eyes closed resting state activity (three min) and eyes open task-related activity (three min). 6.2 Methods Participants 20 children with ADHD participated in the study. All children fulfilled DSM-IV1criteria for ADHD (American Psychiatric Association, 1994). Diagnoses were based on a semi-structured interview with the parents using the Structured Developmental History of the BASC (Reynolds, 2004). WISC-IV (Wechsler, 2003) was administered to estimate IQ. All children were drugfree and without concurring psychotherapy for at least one month before starting the NF training. Children with comorbid neurological or psychiatric disorders, or IQ <80 were excluded from the study. Three children did not complete the study, thus the final sample consisted of 17 children (mean ±SD age: 11.8±2.2 years, one girl). Seven children were diagnosed with inattentive type, ten with combined type. Families were informed about the study from local professionals in the city of Zaragoza (Spain). The experimental design was approved by the Ethical Review Board of the regional health authority and followed the Declaration of Helsinki. Parents and children signed informed consent. Experimental design An open-label pilot study was designed (Figure 6.1). After an intake evaluation, an initial and final behavioral assessment and EEG screening were carried out. The NF training was composed of 18 sessions, executed for two 1DSM was recently updated to the fifth edition (DSM-5). 72 Chapter 6. The effect of NF in ADHD children months (two or three sessions per week). Each session was composed of five trials of four min each for a total of 20 min of training, and a pre- and post- EEG screening. For each EEG screening we recorded three-min of eyes closed resting state activity and three-min of eyes open task-related activity. In the latter, children faced a computer screen showing a square that changed saturation color randomly from gray to red or blue (gradually). Children were instructed to count the number of saturation changes from gray to red as a cognitive challenge (Zoefel et al., 2011). Behavioral assessments Parents rated the clinical conditions of the children pre- and postthe training using the following scales: (i) Parent Rating Scales of the BASC (BASCPRS, Reynolds, 2004). The scores were the externalizing and internalizing problems, and adaptive skills. (ii) Conners’ Parent Rating Scales-Revised (CPRS-R, Conners et al., 1998). The scores were the global index and DSMIV items (inattention, hyperactivity/impulsivity and total score). A battery of neuropsychological tests was administered to the children: (i) Two tests of the WISC-IV (Wechsler, 2003) evaluated working memory. Digit span consisted of sequences of numbers that had to be repeated, either in same or reverse order. Letter-number sequencing consisted of sequences of letters and numbers that had to be repeated in both numerical and alphabetical order. The test scores were the number of correct responses. (ii) D2 test (Brickenkamp and Zillmer, 1998) evaluated focussed and selective attention. Children crossed out target letters on a working sheet, working line by line with 20 s for finishing each line. The score was the concentration index. (iii) Conners’ continuous performance test (CPT II, Conners and Staff, 2000) is a computerized assessment of attention-related problems. The CPT displayed letters on a computer screen, and children had to press the space bar except when the letter “X” was displayed. The test scores were the number of omission and commission errors. Paired samples t-tests were performed for pre vs post comparisons. EEG recording and neurofeedback procedure EEG data was recorded from 16 electrodes placed at FP1, FP2, AFz, F3, Fz, F4, FCz, C3, Cz, C4, P3, Pz, P4, O1, Oz and O2 (subset of the 10/10 system), with the ground and reference electrodes on FPz and on the left earlobe, respectively. EEG was amplified and digitized using a g.tec amplifier (Guger Technologies, Graz, Austria) at a sampling rate of 256 Hz, power-line notchfiltered at 50 Hz and (0.5-60) Hz band-pass filtered. EEG recording and the 6.4. Discussion 79 2) Hz, showed a power decrease measured in relative (p=.008) and absolute power (p < .001). A power increase appeared in beta2 and beta3, (IAF+8, IAF+18) Hz, in relative (p < .001) and absolute power (p=.01). 6.4 Discussion The current study evaluated a subject-specific NF protocol in children diagnosed with ADHD. Individualized approaches may better cope with the large EEG heterogeneity in ADHD and improve the clinical outcome (Arns et al., 2014). Recent NF studies have followed this direction (Arns et al., 2012, Lansbergen et al., 2011a, Logemann et al., 2010). Please note that in our study, “individualized NF approaches” refers to studies determining the EEG trained parameter according to the EEG activity of the individual rather than using a fixed EEG parameter for all the participants of the study. For instance, Arns et al. (2012) classified the individuals into a set of EEG clusters by a comparison to a normative database, and performed different protocols according to the cluster (e.g., theta/beta, alpha or beta suppression, SMR enhancement). Lansbergen et al. (2011a) and Logemann et al. (2010) performed a theta/beta protocol combined with SMR enhancement, in which the feedback sensors and range of frequency bands were determined by a comparison to a normative database. The NF protocol herein proposed aimed at enhancing the relative upper alpha power in frontro-central sites, individually determined using the individual alpha frequency (IAF) as an anchor point. To the best of the authors knowledge, this is the first NF study evaluating such a protocol in ADHD individuals. In comparison to the aforementioned approaches, this NF protocol does not rely on a normative database comparison and it can address recent concerns of children with slow IAF. For example, Lansbergen et al. (2011b) found that children showing slow IAF may be clustered as an excess of theta activity. In addition to that, the use of an unique NF protocol makes possible to perform an homogenous group-level EEG analysis. On one hand, this protocol has the potential to deal with the excess of absolute theta power, which is the most reliable EEG pattern in ADHD to date (Barry et al., 2003, Snyder and Hall, 2006). Due to the 1/fdistribution of EEG power spectra, we hypothetized stronger effects in slow frequencies (power decrease) and upper alpha (power increase). On the other hand, this protocol builds upon the positive results of alpha-based protocols in cognitive performance, mainly evaluated in healthy users (see Gruzelier (2013) for a review on NF studies on cognitive enhancement). Thus, this NF protocol has the potential to alleviate the cognitive deficits of ADHD individuals. Note that deficits 80 Chapter 6. The effect of NF in ADHD children in executive functioning, including working memory, and response inhibition have been repeatedly reported (Barkley, 1997, Martinussen et al., 2005, Castellanos and Tannock, 2002). EEG analysis An extensive power EEG analysis was conducted. We assessed pre-post study and within-session effects in all sensors in the (≈1-30) Hz frequency range by a cluster-based randomization method, obtaining sensor x frequency maps of the power changes. Furthermore, the learning curve over sessions was assessed in fronto-central and parieto-occipital sites for a set of frequency bands covering the (1-30) Hz range. We believe that the present analyses can offer a clearer insight of the electrophysiological effects rather than traditional analyses only in the trained parameter. Children showed pre-post effects in the trained parameter: relative upper alpha power in fronto-central sites was significantly enhanced after the NF training, measured in task-related activity. An average increase of 13% was found, as well a significant positive learning curve over sessions. In line with these results, Nan et al. (2012) performed a similar NF protocol of relative upper alpha enhancement in healthy users, obtaining a positive learning curve over sessions. We also found a significant absolute upper alpha power enhancement in parieto-occipital sites, and a learning curve over sessions. The increase of absolute upper alpha power in NF literature has been related to improvements (in healthy users) in working memory (Escolano et al., 2011) and visuospatial rotation (Zoefel et al., 2011, Hanslmayr et al., 2005). The pre-post effects in task-related activity were mainly restricted to the upper alpha band, with no significant effects in resting state. The stronger effects in task-related activity illustrate the importance of recording EEG in several conditions to provide additional information of the underlying brain processes. This is in contrast to the common practice to study only the resting state, either eyes closed or eyes open. Note that a correlation analysis was conducted between the behavioral and EEG variables, both in the initial and change scores, but no significant results were found. The within-session effects measured the immediate effects after training in resting state and task-related activity, and the effects during training. To do so, EEG data was collected over sessions, and we compared the EEG screenings recorded immediately before vs after the training trials, and the EEG screenings recorded before vs the EEG during the training trials. A significant decrease in absolute and relative upper alpha power appeared in task-related activity, instead of the expected increase. This may be explained by an alpha “rebound” effect. While that kind of effect had been previously 6.4. Discussion 81 reported in EEG literature mainly related to motor acts in alpha or beta activity (Pfurtscheller and Lopes da Silva, 1999), a recent alpha-based NF study with post-traumatic stress disorder (PTSD) patients reported that effect in alpha activity immediately after a single session of training, pointing to homeostatic/compensatory brain mechanisms (Kluetsch et al., 2013). Regarding the resting state, an absolute power decrease was found in slowfrequency oscillations (delta and theta) and lower alpha, as well as a power decrease in lower part of beta. No significant effects in the trained parameter were found during training, however an absolute and relative power decrease appeared in slow-frequency oscillations and lower alpha, as well as an increase in upper part of beta. Thus, although the children were not able to increase the relative upper alpha during training, they showed a strong effect in slow frequencies (as hypothetized) and in upper part of beta to a lower extent. The latter effect was unexpected and should be further explored in future studies. Behavioral assessments Parents reported a significant reduction in the clinical symptoms of the children after the NF training. The externalizing and internalizing problems scores in the BASC test showed a significant improvement, as well as the inattention and hyperactivity/impulsivity scores in the CPRS test. The effect sizes in inattention and hyperactivity/impulsivity were 1.16 and 1.15, respectively. Regarding the neurophysiological tests, children showed a significant improvement in working memory as measured by the digit span and letter-number sequencing tests of the WISC-IV. We found a significant improvement in concentration as assessed by the D2 test. The number of commission errors in the CPT test was significantly decreased, thus suggesting an improvement in impulsivity. No significant change in the number of omission errors was found. Interesting, slightly superior effect sizes in hyperactivity/impulsivity (d= 1.15) were found in comparison to literature (see Arns et al. (2009) meta-analysis). In this direction, a large body of research has hypothetized that the neuronal substrates of inhibitory mechanisms are related to alpha oscillations (Sauseng et al., 2009, Freunberger et al., 2011, Klimesch et al., 2007). Although it should be interpreted with caution, the upper alpha power enhancement herein reported may target mechanisms of behavioral inhibition, thus leading to higher outcomes in hyperactivity/impulsivity symptoms. SMR enhancement also has been traditionally hypothetized to alleviate hyperactivity. Due to the similitudes between these two protocols, the aforementioned relation may account for the clinical improvements and cog- 82 Chapter 6. The effect of NF in ADHD children nitive enhancement in ADHD. It was already pointed out by Hanslmayr et al. (2005) that the results in cognitive enhancement obtained after SMR-based NF (in healthy users) might be in part influenced by upper alpha activity. However, certainly more research is needed to elucidate the mechanisms of action underlying this protocol. Limitations Due to the novelty of the NF protocol in ADHD individuals an open-label pilot study was designed. The number of NF sessions was small in comparison with ADHD literature (30 to 40 sessions are usually executed). Furthermore, non-specific effects of the treatment cannot be ruled out due to the lack of a control group. The positive results of this NF protocol suggest that it should be further explored in a controlled study with a higher number of sessions and a larger sample size. Note that diagnosis was based on DSMIV (American Psychiatric Association, 1994) since it was the more recent edition at the beginning of our study, however it was recently updated to the fifth edition (DSM-5, American Psychiatric Association, 2013). 6.5 Conclusions This study showed that ADHD children can enhance relative and absolute upper alpha activity through NF training (preliminary uncontrolled study). These effects were mainly restricted to upper alpha. Within-session analysis showed a power decrease(“rebound” effect) in task-related activity, with no significant effects during training. In addition, an absolute power decrease appeared in slow-frequency oscillations (note that ADHD children commonly show an excess of slow-frequency activity). Regarding the behavioral measurements, parents rated a clinical improvement in children regarding inattention and hyperactivity/impulsivity, and neurophysiological tests showed an improvement in working memory, concentration and impulsivity (decreased number of commission errors in a continuous performance test). These results suggest that the current protocol is effective in improving several measures of clinical outcome and cognitive performance in ADHD, thus a controlled evaluation seems warranted. 7|Conclusions This thesis has addressed the design of a subject-specific NF approach in a unified framework, and has shown its feasibility by implementing three different NF studies of upper alpha up-regulation for cognitive enhancement in healthy subjects, depressed patients and ADHD children. 7.1 A subject-specific NF approach In order to design of a subject-specific NF approach, we first identified some “methodological principles” applied in the field of BCI in terms of signal processing methods. Note that BCIs have extensively make use of individualized methods for each subject and time of use of the technology. We consider that the main issues that a subject-specific NF approach should take into account are the artifact filtering, the individualization of the EEG brain patterns of interest (both among subjects and over sessions) to deal with the inherent large inter-subject and inter-session variability, and the development of postanalysis methods that take into account these individualized EEG patterns. Below we describe an overview of the state of the art concerning these issues (within NF techniques) and the solutions developed in this thesis. Since we implemented a NF protocol of individual upper alpha up-regulation, these solutions are oriented towards alpha activity. However it is important to keep in mind that this approach could be adapted to other brain patterns, although this may require different implementations. Artifact filtering. Few NF techniques deal with the artifact filtering. For example, Zoefel et al. (2011) performed a threshold-based detection in time domain, stopping the feedback when a blinking artifact was detected. This may reduce the effective time of training. Other techniques include low and high frequency oscillations as additional feedback so the subjects can potentially learn to reduce them (Kober et al., 2014). This may make the training more complex for the subjects and seems not well suited for some types of 83 84 Chapter 7. Conclusions artifacts (e.g., eyes blinking is a natural body system). We developed a realtime artifact filtering method for the blinking artifacts based on Independent Component Analysis (ICA, Hyvarinen, 1999). Note that eyes blinking is a natural body system with high occurrence rate and may distort the alpha activity, specially in frontal electrodes (Delorme et al., 2007). This filter is re-computed per subject and session to deal with the different spatial distribution of this blinking (inter-subject variability) and, for example, to deal with the slight changes in electrode positions over sessions (inter-session variability). Thus this filter might favor the learning of the self-regulation process by proving the subjects with more accurate feedback and/or more time of effective training. Finally, this filter might be of special interest for NF studies targeting frontal locations such as our study with ADHD children. Individualization of brain patterns. The signal processing methods should consider subject-specific EEG brain patterns. In addition, these methods should adapt to the temporal evolution of the brain patterns of each subject, which is commonly implemented in NF literature as a baseline computation before online training. While many NF techniques re-compute the baseline activity of the rhythms over the sessions (Zoefel et al., 2011, Nan et al., 2012, Kober et al., 2013), there is a trend in alpha-based NF procedures in recent years towards the individualization of the alpha frequency by using the Individual Alpha Frequency (IAF, Klimesch, 1999). Some of them compute IAF in resting state conditions (Hanslmayr et al., 2005, Nan et al., 2012, Alexeeva et al., 2012), which may not take into account the inter-task IAF variability (Haegens et al., 2014). Others compute it in task-related activity (Zoefel et al., 2011), which may not be accurate in some subjects due to the well-known alpha attenuation phenomenon during active brain processing. We developed a novel procedure combining resting state and task-related activity which might better deal with the inter-subject variability. We also re-computed the baseline working level of individual upper alpha for each session, measured in task-related activity, in order to accommodate to the temporal evolution of alpha (e.g., dependent on psychological state, outcome of the self-regulation process). Post-analysis of the effects on the EEG. Due to the individualization of brain patterns the developed post-analysis (applied to the EEG data of the three NF studies) measured the effects on the subject-specific brain patterns in both resting state and task-related activity, systematically investigating the effects after the study (pre-post study effects), the immediate effects after the training trials (within-session effects) and the effects over the sessions 7.2. NF studies for cognitive enhancement 85 (trend or learning curve over sessions). Note that a recent review encouraged the use of these metrics (Gruzelier, 2014). In addition to this, we performed an extensive evaluation on the EEG by measuring the effects in non-trained patters (specificity of the training), the effects during the execution of a mental rotation task (transfer effects) and the effects at brain source level. We below outline some methods used throughout the manuscript in the different NF studies. We proposed to use an offline artifact filtering method based on Riemannian geometry (Barachant et al., 2012, 2013). Following this method, the EEG data (of a given session) can be filtered from artifacts by manually selecting a small of artifact-free epochs (15- 20 epochs). This may be a robust method which seems less time-consuming than traditional visual inspection in clinical practice. We proposed the use of a cluster-based randomization procedure to assess the EEG power changes in all the spectral range and sensors, obtaining sensor-frequency maps of the power changes (Nichols and Holmes, 2002, Maris and Oostenveld, 2007). Note that traditional NF studies assess the effects of the training in a small number of pre-determined frequency bands (Zoefel et al., 2011, Nan et al., 2012). While this analysis has been extensively used in BCIs to obtain timefrequency maps (e.g., in motor imagery tasks), its application in NF could provide insights of the specificity of the training in the electrophysiology. We used sLORETA to study the effects at brain source level (Pascual-Marqui, 2002, 2007). The effects at brain source level are not typically assessed, however they might be of special interest for some clinical disorders in which abnormal regulation of brain structures is known, for example the ACC disregulation in depressed patients (Pizzagalli et al., 2001). 7.2 NF studies for cognitive enhancement We implemented a subject-specific NF protocol of upper alpha power upregulation for cognitive enhancement, and performed three studies including healthy and clinical population: patients with major depressive disorder and children with ADHD. We were interested in evaluating whether upper alpha power could be enhanced by means of NF training, and if so, whether it could lead to either enhanced cognition or improvements in clinical outcome. Below we summarize the studies with the main results and contributions. NF study on healthy subjects This study investigated the NF effects on healthy participants following a single session of training (25 minutes) and controlling for non-specific factors (double-blind sham-controlled study). A 86 Chapter 7. Conclusions NF protocol of individual upper alpha up-regulation over parieto-occipital locations was performed. Ten participants were assigned to the experimental group and nine to a sham-feedback control group. We assessed immediate effects after the training and one-day lasting effects on the EEG. In addition, we assessed the changes in the EEG power time-course between the pre- and postexecution of a mental rotation task. The objective of the latter task was to investigate whether the NF training increased pre-stimulus upper alpha power (increasing the desynchronization as well), which is suggested to be related to cognitive performance (Klimesch et al., 2007). These metrics were computed in the trained parameter (upper alpha power) as well as in the (IAF-4, IAF+4) Hz range, thus covering lower alpha 1, (IAF-4, IAF-2) Hz; lower alpha 2, (IAF-2, IAF); and lower beta (IAF+2, IAF+4) Hz. Only the experimental group showed increased upper alpha power (in task-related activity) immediately after the training, as well as increased lower alpha 2 power. The stronger effects appeared in posterior areas of the scalp. Resting state activity was not significantly modified for any group, which suggests that this protocol presents lower effects on resting state. In addition to that, the experimental group presented higher pre-stimulus upper alpha power during the postexecution of the mental rotation task, and consequently higher desynchronization. In relation to other single-session NF studies, Hanslmayr et al. (2005) showed increased pre-stimulus during the execution of a similar mental rotation task after one training session. Ros et al. (2010) also showed EEG changes after alpha-based training over the sensorimotor cortex and an increase in corticospinal excitability. Taken together, these studies suggest that one session of NF training can produce significant effects at electrophysiological level immediately after training. Behavioral analysis showed a more prominent improvement for the experimental group in some cognitive functions (working memory, episodic memory and executive functions), however this improvement was not superior to the one observed for the control group, with the exception of executive functions, which was significantly improved for the experimental group. These results might be explained by a strong learning effect due to repeated measurements and by the short duration of the training. A higher number of training sessions seems thus necessary to yield significant behavioral differences between groups and to increase the effects on the EEG. NF study on depressed patients This study investigated the NF effects on patients with major depressive disorder (MDD). A NF protocol of individual upper alpha up-regulation over parieto-occipital locations was performed. 40 patients were assigned to a eight-session training program and 7.2. NF studies for cognitive enhancement 87 were compared to 20 participants in a non-interventional group. A power EEG analysis and an alpha asymmetry analysis were conducted at the sensor level. sLORETA was used to assess the effect at brain source level. Correlation analysis between the clinical/cognitive and EEG measurements was conducted at both the sensor and brain source level. The experimental group showed increased upper alpha power pre-post study, measured in task-related activity. These effects were not spatially or spectrally restricted to the trained parameter. We extended the EEG power analysis to all sensors in the (1-30 Hz) frequency range by a clustering analysis (Maris and Oostenveld, 2007), obtaining sensor-frequency maps of the power changes. These maps showed significant clusters for the experimental group only, with a power increase after the NF training. The strongest effect in resting state activity appeared in the lower part of alpha (8-10 Hz). Taskrelated activity showed stronger effects in the (4.5-20 Hz) range, apparent in anterior, central and posterior locations. Regarding the effects at brain source level, a current density increase appeared in the alpha band (8-12 Hz) for the experimental group, localized in the subgenual anterior cingulate cortex (sgACC, BA 25). A positive correlation was found for the experimental group between the improvement in processing speed and the increase of beta power at both the sensor and brain source level. Regarding the behavioral effects, we found an improvement in working memory performance and processing speed for the experimental group in comparison to the control group, showing medium-large effect sizes. The objective of this study was to evaluate whether this NF protocol could improve working memory performance in depressed patients, thus alleviating the cognitive deficits of these patients (Gotlib and Joormann, 2010, Castaneda et al., 2008, Austin et al., 2001). To the best of our knowledge, this is the first NF study exploring the cognitive effect of working memory entrainment in depressed patients. Note that recent evidences suggest that the working memory deficits in depression may not only be correlates of depression but also increase the vulnerability and recurrence to depression (Gotlib and Joormann, 2010, Levens and Gotlib, 2010). Thus this NF protocol has potential to improve depressive symptoms. NF study on ADHD children This study investigated the NF effects on children diagnosed with ADHD. We performed a NF protocol of relative upper alpha power enhancement over fronto-central locations. 20 children underwent 18 training sessions (preliminary uncontrolled study). A power EEG analysis assessed pre-post study and within-session effects. Learning curves over sessions were assessed as well. Clinical and neurophysiological 88 Chapter 7. Conclusions variables were measured pre- and posttraining. Children showed increased relative and absolute upper alpha power prepost study, measured in task-related activity, as well as a positive learning curve over sessions. The analysis of within-session effects showed an upper alpha power decrease (“rebound” effect) in task-related activity, similarly to other NF studies (Kluetsch et al., 2013), with no significant effects during training. We further assessed pre-post and within-session power changes in all sensors in the (1-30 Hz) frequency range by a clustering analysis, in both absolute and relative power. While the pre-post effects were mainly restricted to upper alpha band, within-session analysis showed an absolute power decrease in slow-frequency oscillations (delta and theta) in both resting state and task-related activity. Parents rated a clinical improvement in children regarding inattention and hyperactivity/impulsivity. Neurophysiological tests showed an improvement in working memory, concentration and impulsivity (decreased number of commission errors in a continuous performance test). This protocol seems effective in improving several measures of clinical outcome and cognitive performance for ADHD. This is the first NF study evaluating such a protocol in ADHD. This protocol has the potential to deal with the excess of absolute theta power (Barry et al., 2003, Snyder and Hall, 2006) and to target cognitive performance. Note that ADHD individuals show cognitive deficits (Barkley, 1997, Martinussen et al., 2005, Castellanos and Tannock, 2002). This study is in line with individualized NF protocols in children diagnosed with ADHD, which are proposed to better cope with the large EEG heterogeneity in ADHD with regard to traditional/standard approaches (Arns et al., 2012, Lansbergen et al., 2011a, Logemann et al., 2010). A controlled evaluation seems warranted due to the positive results obtained in the current study. 7.3 Summary & future work The results show that the subject-specific NF technique was able to accommodate the large variability of the brain patterns. One important aspect of the NF framework developed is that all the steps of the procedure are fully automatic, thus this approach might be used by non-expert personnel in home settings. This fact could also be used to easily implement double-blind experimental designs. This framework could be seen as a personalized approach in line with the current trend of personalized medicine. It is important to keep in mind that this approach could be adapted towards other brain patterns (different than alpha) which thus may require a different implementation. For example, one might think about comparing the EEG power spectra of an in- Bibliography 95 Congedo, M. (2002). Eureka!(version 3.0)[computer software]. Knoxville, TN: NovaTech EEG Inc. Freeware available at: www. NovaTechEEG. Congedo, M., Lubar, J. F., and Joffe, D. (2004). Low-resolution electromagnetic tomography neurofeedback. Neural Systems and Rehabilitation Engineering, IEEE Transactions on, 12(4):387–397. Conners, C. K., Sitarenios, G., Parker, J. D., and Epstein, J. N. (1998). The revised Conners’ Parent Rating Scale (CPRS-R): factor structure, reliability, and criterion validity. Journal of abnormal child psychology, 26(4):257–268. Conners, C. K. and Staff, M. (2000). Conners’ Continuous Performance Test II (CPT II V. 5). North Tonawanda, NY: Multi-Health Systems Inc. Davidson, R. J. (1998). Anterior electrophysiological asymmetries, emotion, and depression: Conceptual and methodological conundrums. Psychophysiology, 35(05):607–614. Davidson, R. J. (2004). What does the prefrontal cortex “do” in affect: perspectives on frontal EEG asymmetry research. Biological psychology, 67(1):219–234. Davidson, R. J., Pizzagalli, D., Nitschke, J. B., and Putnam, K. (2002). Depression: perspectives from affective neuroscience. Annual review of psychology, 53(1):545–574. Delorme, A., Sejnowski, T., and Makeig, S. (2007). Enhanced detection of artifacts in EEG data using higher-order statistics and independent component analysis. Neuroimage, 34(4):1443–1449. Demos, J. N. (2005). Getting started with neurofeedback. WW Norton & Company. DeRubeis, R., Siegle, G., and Hollon, S. (2008). Cognitive therapy versus medication for depression: treatment outcomes and neural mechanisms. Nat. Rev. Neurosci., 9(10):788–796. Engel, A. K. and Fries, P. (2010). Beta-band oscillations-signalling the status quo? Current opinion in neurobiology, 20(2):156–165. Enriquez-Geppert, S., Huster, R. J., and Herrmann, C. S. (2013). Boosting brain functions: improving executive functions with behavioral training, neurostimulation, and neurofeedback. International Journal of Psychophysiology, 88(1):1–16. 96 Bibliography Enriquez-Geppert, S., Huster, R. J., Scharfenort, R., Mokom, Z. N., Zimmermann, J., and Herrmann, C. S. (2014). Modulation of frontal-midline theta by neurofeedback. Biological psychology, 95:59–69. Escolano, C., Aguilar, M., Lopez, M., and Minguez, J. (2014a). Sistema y método de entrenamiento cognitivo mediante neuromodulación. WO Patent App. PCT/ES2013/070,781. Escolano, C., Aguilar, M., and Minguez, J. (2011). EEG-based upper alpha neurofeedback training improves working memory performance. In International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pages 2327–2330, Boston (USA). Escolano, C., Antelis, J. M., and Minguez, J. (2012a). A telepresence mobile robot controlled with a non-invasive brain-computer interface. IEEE Transactions on Systems, Man, and Cybernetics—Part B: Cybernetics, 42(3):793–804. Escolano, C. and Minguez, J. (2011). Sistema de teleoperación multi-robot basado en interfaz cerebro-computador. Revista Iberoamericana de Automática e Informática Industrial (RIAI), 8(2):16–23. Escolano, C., Navarro-Gil, M., Garcia-Campayo, J., Congedo, M., De Ridder, D., and Minguez, J. (2014b). A controlled study on the cognitive effect of alpha neurofeedback training in patients with major depressive disorder. Frontiers in Behavioral Neuroscience, 8(296). Escolano, C., Navarro-Gil, M., Garcia-Campayo, J., Congedo, M., and Minguez, J. (2014c). The effects of individual upper alpha neurofeedback in ADHD: An open-label pilot study. Applied Psychophysiology and Biofeedback (In Press). Escolano, C., Navarro-Gil, M., Garcia-Campayo, J., and Minguez, J. (2013). EEG-based upper-alpha neurofeedback for cognitive enhancement in major depressive disorder: A preliminary, uncontrolled study. In International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pages 6293–6296, Osaka (Japan). Escolano, C., Navarro-Gil, M., Garcia-Campayo, J., and Minguez, J. (2014d). The effects of a single session of upper alpha neurofeedback for cognitive enhancement: A sham-controlled study. Applied Psychophysiology and Biofeedback (In Press). Bibliography 97 Escolano, C., Olivan, B., Lopez-del-Hoyo, Y., Garcia-Campayo, J., and Minguez, J. (2012b). Double-blind single-session neurofeedback training in upper-alpha for cognitive enhancement of healthy subjects. In International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pages 4643–4647, San Diego (USA). Escolano, C., Ramos, A., Matuz, T., Birbaumer, N., and Minguez, J. (2010). A telepresence robotic system operated with a P300-based brain-computer interface: Initial tests with ALS patients. In International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pages 4476–4480, Buenos Aires (Argentina). Faraone, S. V., Biederman, J., and Mick, E. (2006). The age-dependent decline of attention deficit hyperactivity disorder: a meta-analysis of followup studies. Psychological medicine, 36(2):159–166. Freunberger, R., Werkle-Bergner, M., Griesmayr, B., Lindenberger, U., and Klimesch, W. (2011). Brain oscillatory correlates of working memory constraints. Brain research, 1375:93–102. Goljahani, A., D’Avanzo, C., Schiff, S., Amodio, P., Bisiacchi, P., and Sparacino, G. (2012). A novel method for the determination of the EEG individual alpha frequency. NeuroImage, 60(1):774–786. Gotlib, I. H. (1998). EEG alpha asymmetry, depression, and cognitive functioning. Cognition & Emotion, 12(3):449–478. Gotlib, I. H. and Joormann, J. (2010). Cognition and depression: current status and future directions. Annual review of clinical psychology, 6:285. Graham, J., Banaschewski, T., Buitelaar, J., Coghill, D., Danckaerts, M., Dittmann, R., Döpfner, M., Hamilton, R., Hollis, C., Holtmann, M., et al. (2011). European guidelines on managing adverse effects of medication for ADHD. European child & adolescent psychiatry, 20(1):17–37. Gronwall, D. M. A. (1977). Paced auditory serial-addition task: a measure of recovery from concussion. Perceptual and motor skills, 44(2):367–373. Gruzelier, J. H. (2013). EEG-neurofeedback for optimising performance. I: A review of cognitive and affective outcome in healthy participants. Neuroscience & Biobehavioral Reviews, 44:124–141. Gruzelier, J. H. (2014). EEG-neurofeedback for optimising performance. III: A review of methodological and theoretical considerations. Neuroscience & Biobehavioral Reviews, 44:159–182. 98 Bibliography Haegens, S., Cousijn, H., Wallis, G., Harrison, P. J., and Nobre, A. C. (2014). Inter-and intra-individual variability in alpha peak frequency. NeuroImage, 92:46–55. Hagemann, D., Naumann, E., and Thayer, J. F. (2001). The quest for the EEG reference revisited: A glance from brain asymmetry research. Psychophysiology, 38(05):847–857. Hammond, D. (2005). Neurofeedback treatment of depression and anxiety. Journal of Adult Development, 12(2):131–137. Hammond, D. C. (2000). Neurofeedback treatment of depression with the Roshi. Journal of Neurotherapy, 4(2):45–56. Hanslmayr, S., Sauseng, P., Doppelmayr, M., Schabus, M., and Klimesch, W. (2005). Increasing individual upper alpha power by neurofeedback improves cognitive performance in human subjects. Applied Psychophysiology and Biofeedback, 30:1–10. Heinrich, H., Gevensleben, H., and Strehl, U. (2007). Annotation: Neurofeedback–train your brain to train behaviour. Journal of Child Psychology and Psychiatry, 48(1):3–16. Henriques, J. B. and Davidson, R. J. (1991). Left frontal hypoactivation in depression. Journal of abnormal psychology, 100(4):535. Holmes, A. P., Blair, R., Watson, G., and Ford, I. (1996). Nonparametric analysis of statistic images from functional mapping experiments. Journal of Cerebral Blood Flow & Metabolism, 16(1):7–22. Holroyd, C. B., Krigolson, O. E., Baker, R., Lee, S., and Gibson, J. (2009). When is an error not a prediction error? An electrophysiological investigation. Cognitive, Affective, & Behavioral Neuroscience, 9(1):59–70. Hughes, J. R. and John, E. R. (1999). Conventional and quantitative electroencephalography in psychiatry. The Journal of neuropsychiatry and clinical neurosciences, 11(2):190–208. Hyvarinen, A. (1999). Fast and robust fixed-point algorithms for independent component analysis. Neural Networks, IEEE Transactions on, 10(3):626– 634. Iturrate, I., Antelis, J. M., Kubler, A., and Minguez, J. (2009). A noninvasive brain-actuated wheelchair based on a P300 neurophysiological protocol and automated navigation. Robotics, IEEE Transactions on, 25(3):614–627. Bibliography 99 Jensen, O., Gelfand, J., Kounios, J., and Lisman, J. E. (2002). Oscillations in the alpha band (9–12 Hz) increase with memory load during retention in a short-term memory task. Cerebral Cortex, 12(8):877–882. Jensen, O. and Mazaheri, A. (2010). Shaping functional architecture by oscillatory alpha activity: gating by inhibition. Frontiers in human neuroscience, 4. Kamiya, J. (1969). Operant control of the EEG alpha rhythm and some of its reported effects on consciousness. Altered states of consciousness. New York: Wiley, 1069. Keizer, A. W., Verment, R. S., and Hommel, B. (2010). Enhancing cognitive control through neurofeedback: A role of gamma-band activity in managing episodic retrieval. Neuroimage, 49(4):3404–3413. Kemp, A., Griffiths, K., Felmingham, K., Shankman, S., Drinkenburg, W., Arns, M., Clark, C., and Bryant, R. (2010). Disorder specificity despite comorbidity: Resting EEG alpha asymmetry in major depressive disorder and post-traumatic stress disorder. Biological Psychology, 85(2):350–354. Kessler, R. C., Berglund, P., Demler, O., Jin, R., Koretz, D., Merikangas, K. R., Rush, A. J., Walters, E. E., and Wang, P. S. (2003). The epidemiology of major depressive disorder. JAMA: the journal of the American Medical Association, 289(23):3095–3105. Klimesch, W. (1999). EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. Brain Research Reviews, 29(2-3):169 – 195. Klimesch, W., Sauseng, P., and Gerloff, C. (2003). Enhancing cognitive performance with repetitive transcranial magnetic stimulation at human individual alpha frequency. European Journal of Neuroscience, 17(5):1129– 1133. Klimesch, W., Sauseng, P., and Hanslmayr, S. (2007). EEG alpha oscillations: the inhibition-timing hypothesis. Brain research reviews, 53(1):63– 88. Klimesch, W., Schimke, H., and Pfurtscheller, G. (1993). Alpha frequency, cognitive load and memory performance. Brain topography, 5(3):241–251. Kluetsch, R., Ros, T., Théberge, J., Frewen, P., Calhoun, V., Schmahl, C., Jetly, R., and Lanius, R. (2013). Plastic modulation of PTSD resting-state 100 Bibliography networks and subjective wellbeing by EEG neurofeedback. Acta Psychiatrica Scandinavica. Kober, S. E., Witte, M., Ninaus, M., Neuper, C., and Wood, G. (2013). Learning to modulate one’s own brain activity: the effect of spontaneous mental strategies. Frontiers in Human Neuroscience, 7:695. Kober, S. E., Witte, M., Stangl, M., Väljamäe, A., Neuper, C., and Wood, G. (2014). Shutting down sensorimotor interference unblocks the networks for stimulus processing: An SMR neurofeedback training study. Clinical Neurophysiology. Kroenke, K., Spitzer, R. L., and Williams, J. B. (2001). The PHQ-9. Journal of general internal medicine, 16(9):606–613. La Vaque, T. J. and Rossiter, T. (2001). The ethical use of placebo controls in clinical research: the Declaration of Helsinki. Applied psychophysiology and biofeedback, 26(1):23–37. Lancaster, J. L., Woldorff, M. G., Parsons, L. M., Liotti, M., Freitas, C. S., Rainey, L., Kochunov, P. V., Nickerson, D., Mikiten, S. A., and Fox, P. T. (2000). Automated Talairach atlas labels for functional brain mapping. Human brain mapping, 10(3):120–131. Lansbergen, M., van Dongen-Boomsma, M., Buitelaar, J. K., and Slaats- Willemse, D. (2011a). ADHD and EEG-neurofeedback: a double-blind randomized placebo-controlled feasibility study. Journal of Neural Transmission, 118(2):275–284. Lansbergen, M. M., Arns, M., van Dongen-Boomsma, M., Spronk, D., and Buitelaar, J. K. (2011b). The increase in theta/beta ratio on resting-state EEG in boys with attention-deficit/hyperactivity disorder is mediated by slow alpha peak frequency. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 35(1):47–52. Lee, T. W., Girolami, M., and Sejnowski, T. J. (1999). Independent component analysis using an extended infomax algorithm for mixed subgaussian and supergaussian sources. Neural computation, 11(2):417–441. Levens, S. M. and Gotlib, I. H. (2010). Updating positive and negative stimuli in working memory in depression. Journal of experimental psychology: general, 139(4):654. Bibliography 101 Lezak, M. D. (2004). Neuropsychological assessment. Oxford University Press, New York. Logemann, H. N., Lansbergen, M. M., Van Os, T. W., Böcker, K. B., and Kenemans, J. L. (2010). The effectiveness of EEG-feedback on attention, impulsivity and EEG: a sham feedback controlled study. Neuroscience letters, 479(1):49–53. Loo, S. K. and Makeig, S. (2012). Clinical utility of EEG in attentiondeficit/hyperactivity disorder: a research update. Neurotherapeutics, 9(3):569–587. Lubar, J. F. (1991). Discourse on the development of EEG diagnostics and biofeedback for attention-deficit/hyperactivity disorders. Biofeedback and Self-regulation, 16(3):201–225. Lubar, J. F., Congedo, M., and Askew, J. H. (2003). Low-resolution electromagnetic tomography (LORETA) of cerebral activity in chronic depressive disorder. International Journal of Psychophysiology, 49(3):175–185. Lubar, J. F. and Shouse, M. N. (1976). EEG and behavioral changes in a hyperkinetic child concurrent with training of the sensorimotor rhythm (SMR). Biofeedback and Self-regulation, 1(3):293–306. Maris, E. and Oostenveld, R. (2007). Nonparametric statistical testing of EEG- and MEG-data. Journal of neuroscience methods, 164(1):177–90. Martinussen, R., Hayden, J., Hogg-Johnson, S., and Tannock, R. (2005). A meta-analysis of working memory impairments in children with attentiondeficit/hyperactivity disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 44(4):377–384. Mathers, C. D. and Loncar, D. (2006). Projections of global mortality and burden of disease from 2002 to 2030. PLoS medicine, 3(11):e442. Mayberg, H. S. (1997). Limbic-cortical dysregulation: a proposed model of depression. The Journal of neuropsychiatry and clinical neurosciences. Mayberg, H. S., Liotti, M., Brannan, S. K., McGinnis, S., Mahurin, R. K., Jerabek, P. A., Silva, J. A., Tekell, J. L., Martin, C. C., Lancaster, J. L., et al. (1999). Reciprocal limbic-cortical function and negative mood: converging PET findings in depression and normal sadness. American Journal of Psychiatry, 156(5):675–682. 102 Bibliography Mazziotta, J., Toga, A., Evans, A., Fox, P., Lancaster, J., Zilles, K., Woods, R., Paus, T., Simpson, G., Pike, B., et al. (2001). A probabilistic atlas and reference system for the human brain: International Consortium for Brain Mapping (ICBM). Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences, 356(1412):1293–1322. McNab, F. and Klingberg, T. (2007). Prefrontal cortex and basal ganglia control access to working memory. Nature neuroscience, 11(1):103–107. Millán, J. d. R., Rupp, R., Müller-Putz, G. R., Murray-Smith, R., Giugliemma, C., Tangermann, M., Vidaurre, C., Cincotti, F., Kübler, A., Leeb, R., et al. (2010). Combining brain–computer interfaces and assistive technologies: state-of-the-art and challenges. Frontiers in neuroscience, 4. Miranda, J. P. and Valencia, R. R. (1997). English and Spanish versions of a memory test: Word-length effects versus spoken-duration effects. Hispanic J. of Behavioral Sciences, 19(2):171–181. Molina, B. S., Hinshaw, S. P., Swanson, J. M., Arnold, L. E., Vitiello, B., Jensen, P. S., Epstein, J. N., Hoza, B., Hechtman, L., Abikoff, H. B., et al. (2009). The MTA at 8 years: prospective follow-up of children treated for combined-type ADHD in a multisite study. Journal of the American Academy of Child & Adolescent Psychiatry, 48(5):484–500. Monastra, V. J., Lynn, S., Linden, M., Lubar, J. F., Gruzelier, J., and La Vaque, T. J. (2006). Electroencephalographic biofeedback in the treatment of attention-deficit/hyperactivity disorder. Journal of Neurotherapy, 9(4):5–34. Nan, W., Rodrigues, J. P., Ma, J., Qu, X., Wan, F., Mak, P. I., Mak, P. U., Vai, M. I., and Rosa, A. (2012). Individual alpha neurofeedback training effect on short term memory. International Journal of Psychophysiology, 86(1):83–87. Neuper, C., Scherer, R., Reiner, M., and Pfurtscheller, G. (2005). Imagery of motor actions: Differential effects of kinesthetic and visual–motor mode of imagery in single-trial EEG. Cognitive Brain Research, 25(3):668–677. Nichols, T. E. and Holmes, A. P. (2002). Nonparametric permutation tests for functional neuroimaging: a primer with examples. Human brain mapping, 15(1):1–25. Bibliography 103 Niedermeyer, E. and da Silva, F. L. (2005). Electroencephalography: basic principles, clinical applications, and related fields. Lippincott Williams & Wilkins. Ninaus, M., Kober, S. E., Witte, M., Koschutnig, K., Stangl, M., Neuper, C., and Wood, G. (2013). Neural substrates of cognitive control under the belief of getting neurofeedback training. Frontiers in human neuroscience, 7. Niv, S. (2013). Clinical efficacy and potential mechanisms of neurofeedback. Personality and Individual Differences, 54(6):676 – 686. Palva, S. and Palva, J. M. (2007). New vistas for alpha-frequency band oscillations. Trends in neurosciences, 30(4):150–8. Pascual-Marqui, R. (2002). Standardized low-resolution brain electromagnetic tomography (sLORETA): technical details. Methods Find Exp Clin Pharmacol, 24(Suppl D):5–12. Pascual-Marqui, R. (2007). Discrete, 3D distributed, linear imaging methods of electric neuronal activity. Part 1: exact, zero error localization. arXiv preprint arXiv:0710.3341. Pesarin, F. (2001). Multivariate permutation tests: with applications in biostatistics, volume 240. Wiley, NewYork, NY. Pfurtscheller, G. and Lopes da Silva, F. H. (1999). Event-related EEG/MEG synchronization and desynchronization: basic principles. Clinical Neurophysiology, 110(11):1842–1857. Pfurtscheller, G. and Neuper, C. (2001). Motor imagery and direct braincomputer communication. Proceedings of the IEEE, 89(7):1123–1134. Pfurtscheller, G., Stancak Jr, A., and Neuper, C. (1996). Event-related synchronization (ERS) in the alpha band – an electrophysiological correlate of cortical idling: a review. International Journal of Psychophysiology, 24(1):39–46. Pizzagalli, D., Pascual-Marqui, R. D., Nitschke, J. B., Oakes, T. R., Larson, C. L., Abercrombie, H. C., Schaefer, S. M., Koger, J. V., Benca, R. M., and Davidson, R. J. (2001). Anterior cingulate activity as a predictor of degree of treatment response in major depression: evidence from brain electrical tomography analysis. American Journal of Psychiatry, 158(3):405–415. 104 Bibliography Polanczyk, G., de Lima, M., Horta, B., Biederman, J., and Rohde, L. (2007). The worldwide prevalence of ADHD: a systematic review and metaregression analysis. American journal of psychiatry, 164(6):942–948. Ramos-Murguialday, A., Broetz, D., Rea, M., Läer, L., Yilmaz, Ö., Brasil, F. L., Liberati, G., Curado, M. R., Garcia-Cossio, E., Vyziotis, A., et al. (2013). Brain–machine interface in chronic stroke rehabilitation: A controlled study. Annals of neurology, 74(1):100–108. Reid, S. A., Duke, L. M., and Allen, J. J. (1998). Resting frontal electroencephalographic asymmetry in depression: Inconsistencies suggest the need to identify mediating factors. Psychophysiology, 35(4):389–404. Reitan, R. M. (1958). Validity of the Trail Making Test as an indicator of organic brain damage. Perceptual and motor skills, 8(3):271–276. Rey, A. (1964). L’examen clinique en psychologie. Presses Universitaires de France, Paris. Reynolds, C. R. (2004). Behavior assessment system for children. Wiley Online Library. Ros, T., Munneke, M., Parkinson, L., and Gruzelier, J. (2014). Neurofeedback facilitation of implicit motor learning. Biological Psychology, 95(0):54 – 58. {SAN} (Society of Applied Neuroscience) Special issue on Neurofeedback. Ros, T., Munneke, M. A., Ruge, D., Gruzelier, J. H., and Rothwell, J. C. (2010). Endogenous control of waking brain rhythms induces neuroplasticity in humans. The European journal of neuroscience, 31(4):770–8. Ros, T., Théberge, J., Frewen, P. A., Kluetsch, R., Densmore, M., Calhoun, V. D., and Lanius, R. A. (2013). Mind over chatter: plastic up-regulation of the fMRI salience network directly after EEG neurofeedback. Neuroimage, 65:324–335. Rushworth, M. F. and Behrens, T. E. (2008). Choice, uncertainty and value in prefrontal and cingulate cortex. Nature neuroscience, 11(4):389–397. Salinsky, M., Oken, B., and Morehead, L. (1991). Test-retest reliability in EEG frequency analysis. Electroencephalography and clinical neurophysiology, 79(5):382–392.