Neurofeedback training based on motor imagery strategies increases EEG complexity in elderly population

entropy Article Neurofeedback Training Based on Motor Imagery Strategies Increases EEG Complexity in Elderly Population Diego Marcos-Martínez 1,* , Víctor Martínez-Cagigal 1,2 , Eduardo Santamaría-Vázquez 1,2 , Sergio Pérez-Velasco 1and Roberto Hornero 1,2   Citation: Marcos-Martínez, D.; Martínez-Cagigal, V.; Santamaría-Vázquez, E; Pérez-Velasco, S; Hornero, R. Motor Imagery-Based Neurofeedback Training Increases EEG Complexity in Elderly Population. Entropy 2021,23, 1574. https://doi.org/10.3390/ e23121574 Academic Editor: Carlos M. Travieso-González Received: 21 September 2021 Accepted: 24 November 2021 Published: 25 November 2021 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). 1Biomedical Engineering Group, E.T.S.I. Telecomunicación, University of Valladolid, Paseo de Belén 15, 47011 Valladolid, Spain ; victor[email protected] (V.M.-C.); [email protected] (E.S.-V.); sergio.per[email protected] (S.P.-V.); [email protected] (R.H.) 2Centro de Investigación Biomédica en Red en Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN), 28029 Madrid, Spain *Correspondence: diego.mar[email protected]; Tel.: +34-983-423-708 Abstract: Neurofeedback training (NFT) has shown promising results in recent years as a tool to address the effects of age-related cognitive decline in the elderly. Since previous studies have linked reduced complexity of electroencephalography (EEG) signal to the process of cognitive decline, we propose the use of non-linear methods to characterise changes in EEG complexity induced by NFT. In this study, we analyse the pre- and post-training EEG from 11 elderly subjects who performed an NFT based on motor imagery (MI–NFT). Spectral changes were studied using relative power (RP) from classical frequency bands (delta, theta, alpha, and beta), whilst multiscale entropy (MSE) was applied to assess EEG-induced complexity changes. Furthermore, we analysed the subject’s scores from Luria tests performed before and after MI–NFT. We found that MI–NFT induced a power shift towards rapid frequencies, as well as an increase of EEG complexity in all channels, except for C3. These improvements were most evident in frontal channels. Moreover, results from cognitive tests showed significant enhancement in intellectual and memory functions. Therefore, our findings suggest the usefulness of MI–NFT to improve cognitive functions in the elderly and encourage future studies to use MSE as a metric to characterise EEG changes induced by MI–NFT. Keywords: neurofeedback training (NFT); motor imagery (MI); sample entropy; multiscale entropy (MSE); brain–computer interfaces (BCI); elderly people; age-relate cognitive decline; Luria adult neuropsychological diagnosis (Luria-AND) 1. Introduction Electroencephalography (EEG) is a non-invasive and portable method of monitoring brain activity. This technique is based on recording the electrical activity from pyramidal neurons of the cortex by placing a set of electrodes on the subject’s scalp [ 1 ]. Task-related EEG patterns (e.g., visual stimuli or motor intentions) are used by brain–computer interfaces (BCI) to predict the user’s intentions and convert them into commands to control an external device, without using muscles or peripheral nerves [ 1 , 2 ]. Through this direct communication between the subject’s brain and an external device, BCI applications aim to improve the quality of life of people with motor or cognitive disabilities [ 1 , 2 ]. Nevertheless, assisting the disabled is not the only objective of BCI, but also the rehabilitation or recovery of their motor and cognitive functions [3,4]. Neurofeedback training (NFT) is a therapy based on the hypothesis that, due to brain plasticity, effects of neural disorders can be counteracted by inducing the appropriate brain modulation that normalizes the patient’s deviant brain signal [ 5 , 6 ]. In this regard, NFT users are encouraged to modulate their EEG signals (e.g., the power of a specific frequency band or a ratio between band powers), which is expected to have beneficial effects on their brain state and can lead to brain microstructural changes after training [ 7 ]. Entropy 2021,23, 1574. https://doi.org/10.3390/e23121574 https://www.mdpi.com/journal/entropy Entropy 2021,23, 1574 2 of 19 To this end, BCI applications are employed to measure users’ EEG signals and then provide a feedback stimulus that assists them in finding strategies to gain control of their brain signal modulation. Hereafter, we will refer to this training paradigm as classical NFT. In recent years, several studies have investigated NFT-induced neurological improvements in patients with attention-deficit/hyperactivity disorder [ 8 , 9 ] or epilepsy [ 10 ], among others. Moreover, age-related brain changes have been shown to lead a power shift from rapid to slow rhythms in brain activity [ 11 – 14 ]. Thus, NFT has been proposed as a promising strategy to prevent the progression of the effects of age-related cognitive decline [ 15 ]. Accordingly, NFT aims to normalize the deviated power spectral distribution of the patient in order to enhance their cognitive functions [ 5 ]. In view of increasing life expectancy [ 16 ], NFT could improve the social well-being of the growing elderly population in the future. Previous classical NFT studies on healthy young adults showed significant changes in theta (4–8 Hz) [ 17 , 18 ], alpha (8–13 Hz) [ 19 – 21 ], and beta (13–30 Hz) [ 21 ] band powers. These studies also reported significant improvements in the results of neuropsychological tests for the assessment of memory-related [17,19,21] , attention [ 17 , 21 ], and visuospatial functions [ 20 ]. Furthermore, several studies have been conducted on elderly subjects [ 15 ]. Such studies reported significant differences in the theta [ 17 , 22 , 23 ], alpha [ 22 – 24 ], beta [22,25] , and gamma ( > 30 Hz) [ 25 ] band powers, as well as significant improvements in memory-related functions [17,22–24,26] and attention [17,22–24,26] after the NFT. On the other hand, EEG activity involved in mental motor imagery (MI) tasks is associated with motor and cognitive functions [ 27 , 28 ]. These MI tasks are based on the mental imagination of a movement without any peripheral muscle activation and produce desynchronization and synchronization events in alpha and beta frequency bands over the contralateral sensorimotor areas [ 29 ]. These events are called sensorimotor rhythms (SMR) and can be used as control signals by BCI applications. Hence, some studies have proposed the use of a MI-based NFT paradigm (MI–NFT), instead of classical NFT, as a promising approach to achieve the desired modulation [ 3 , 4 , 30 ]. The MI–NFT paradigm is broadly extended among studies focusing on neurorehabilitation of post-stroke patients [ 3 , 4 , 31 – 35 ]. This paradigm has proven to be effective in promoting functional and structural brain plasticity and recovery of motor function [ 36 ]. Furthermore, in contrast to the classical NFT, the MI–NFT paradigm allows to develop BCI applications with two degrees of freedom (right hand vs. left hand MI). Therefore, the developed training interfaces can be more complex. In this sense, studies have suggested that more gamified applications may be beneficial for therapy outcomes [ 32 , 37 ]. In view of the results achieved by the MI–NFT paradigm in neurorehabilitation studies of stroke patients, it is interesting to consider this paradigm for cognitive training. Indeed, gaining control of SMR modulation may be beneficial for patients, as previous research has suggested an active role of alpha brain activity in cognitive functions, such as memory-related processes [ 38 – 40 ], intelligence [ 41 ], executive functions [ 40 ], and attention-related functions [ 19 , 39 , 42 ], as well as a functional role of beta oscillations in working memory [ 43 , 44 ], language comprehension [ 39 , 45 ], and attention-related functions [ 19 , 44 , 46 ]. In particular, our previous work reported significant EEG changes in alpha and beta bands, as well as an enhancement of visuospatial, language, memory, and intellectual functions in healthy elderly people after 5 MI–NFT sessions [ 30 ]. Despite being a suitable approach to NFT-based cognitive training, there is a lack of research on the impact of the MI–NFT paradigm on users’ brain activity and cognitive functions. Even though the results from NFT-based cognitive training studies are encouraging, as they reveal older people’s ability to control their own brain activity [ 15 ], further analysis of NFT’s effects on the subject’s brain activity is lacking. That is, EEG changes induced by NFT are mainly assessed by linear techniques based on spectral analysis. This may not be enough to determine the overall impact that NFT can produce on the subjects’ brain activity. In this regard, non-linear analysis methods have been proposed as a suitable tool to give insight into brain dynamics due to its non-linear coupling between neuronal populations [ 12 ]. Particularly, multiscale entropy (MSE) was proposed to analyse the complexity of biological signals on multiple time scales [ 47 ]. This metric has proven useful Entropy 2021,23, 1574 3 of 19 for exploring changes in EEG complexity [ 48 – 51 ], which has been shown to be related to cognition state [ 12 , 13 ]. Indeed, a recent study applied for the first time a variation of MSE algorithm to assess the effects of classical NFT on the subject’s EEG complexity, showing higher complexity values after training [ 52 ]. These results encourage further study of the NFT’s influence on brain signal complexity. To the best of our knowledge, changes in EEG complexity induced by MI–NFT in the elderly have not yet been explored. In this sense, we hypothesize that the effects of cognitive training based on the MI–NFT paradigm could also manifest themselves as changes in the complexity of subject’s EEG signal. Thus, differences in complexity between pre- and post-training measures could be expected to be found. To assess these changes, we propose to apply MSE based on sample entropy (SampEn) due to its effectiveness in analysing the irregularity and complexity of biological time-series [ 53 ]. Therefore, the objective of the present study is two-fold: (i) to comprehensively evaluate EEG changes induced in elderly subjects after an MI–NFT and (ii) to analyse the cognitive improvement achieved by the subjects after performing the MI–NFT. 2. Materials and Methods 2.1. Experimental Protocol To assess the changes in brain activity induced by a cognitive training based on MI–NFT, we employed a dataset recorded in our previous work [ 30 ]. A scheme of the experimental protocol is shown in Figure 1. The study involved 63 subjects over 60 years recruited from ’Centro de Referencia Estatal de San Andrés del Rabanedo’ (León, Spain). All subjects were healthy, free of psychotropic medication, and without previous psychiatric or neurological disorders or substance abuse. None of them had previous experience in using BCI. They were randomly divided into a control group (32 subjects) and an NFT group (31 subjects). Participants from both groups carried out a Luria adult neuropsychological diagnosis (Luria-AND) test [54] to analyse their neuropsychological status prior to NFT. Figure 1. Scheme of the experimental protocol of the MI-based NFT study. An example of the feedback from a MI task is shown. In order to perform the training, a novel MI-based BCI tool was developed. Eight active electrodes (F3, F4, T7, C3, Cz, C4, T8, and Pz) were used, placed in an elastic cap according to the international 10-20 system distribution [ 55 ]. Ground electrode was located at AFz channel, whilst the common reference was placed in the earlobe. Training tasks were carried out using a g.USBamp amplifier (Guger Technologies OG, Graz, Austria). EEG signals were acquired at 256 Hz sampling rate and processed in real time using the BCI2000 general-purpose system [ 56 ]. The feedback stimulus was provided using the band power of tree spectral bands centred on 12, 18, and 21 Hz, with bandwidth of 3 Hz. Entropy 2021,23, 1574 4 of 19 The training was performed only by NFT group and consisted of 5 sessions, with an average duration of 90 min over 5 weeks [ 30 ]. Five MI tasks with a different level of difficulty were designed. The MI–NFT paradigm allowed the development of tasks with two degrees of freedom which combine the modulation of SMR and the use of cognitive functions: • Task 1: Subjects were required to imagine hand movements. Alternatively, a closed door or a closed window was displayed on the screen. Then, the paradigm indicated whether users had to perform right- or left-hand MI, respectively. If the user correctly performed the exercise, the displayed object was opened. This task aimed to help subjects learn to modulate their SMR. • Task 2: A target randomly located on the right or the left of the screen was displayed 3 s prior to the start of the trial. In order to reach the target with a displayed cursor, subjects had to continuously perform MI for a maximum duration of 10 s. The target and the cursor were represented by different pairs of related pictures, such as fish/fridge or person/house. • Task 3: Subjects had to control a cursor in order to reach one of the two displayed targets (one of them was a picture related to cursor’s picture, and the other was unrelated to it) by performing MI. Targets were displayed 3 s prior to the start of the trial, and subjects had to complete the task in a maximum time of 10 s. Therefore, this task not only involves MI but also the use of logic. • Task 4: A person walking forward continuously on a path was displayed. Through the MI of their hands, subjects had to control the horizontal position in order to overcome the different obstacles that appeared on the path. This task required the use of the visuospatial function of the subjects. In each trial, the duration of the feedback period was 18 s. • Task 5: Two pairs of images were shown sequentially for 3 s each, so that one image was common to both pairs. Subjects had to identify the image shown twice and move the cursor towards it by performing MI. They had a maximum of 12 s to do so. Thus, this activity also involved the use of memory. The easiest tasks (i.e., Task 1 and Task 2) were performed at first and the difficulty was increased gradually throughout the sessions. Examples of Task 3 and Task 4 are shown in Figure 2. All subjects performed again the Luria-AND test after the training period to assess possible changes in the different neuropsychological functions. A complete description of the experimental protocol and EEG recording procedures can be found in [30]. Figure 2. Examples of training interface. Picture ( A ) shows sequences of screenshots from Task 3. In the upper sequence, the cursor is displayed as fish, so the subject has to perform right hand MI to reach the correct target (fridge on the right), while in the lower sequence, the cursor is displayed as trousers, so the subject has to perform left hand MI to reach the correct target (cupboard on the left). In picture ( B ), an example of Task 4 is depicted. The subject has to perform MI in order to overcome the displayed obstacles in real time. Entropy 2021,23, 1574 5 of 19 2.2. Dataset From the NFT group, pre- and post-training EEG recordings were performed in only 11 subjects (7 females; mean age = 69.4 ± 5.75 years). Therefore, the dataset employed in this study is composed of EEG signals from these subjects. The recordings consisted of two-minute EEG signals in resting state with eyes closed at the beginning of the first session and at the end of the last training session. Detailed socio-demographic information about the 11 NFT subjects under study is shown in Table 1. In addition, the Luria-AND scores of these 11 subjects were used in order to assess their cognitive changes after MI–NFT. A comprehensive analysis of Luria-AND scores of the remaining subjects can be found in [30]. Table 1. Socio-demographic data of the population included in the study. Socio-Demographic Data Identifier Sex Age (Years) U01 Female 70 U02 Female 65 U03 Female 65 U04 Male 71 U05 Male 68 U06 Female 73 U07 Male 81 U08 Male 65 U09 Female 75 U10 Female 70 U11 Female 60 2.3. Luria Adult Neuropsychological Diagnosis Luria-AND [ 54 ] is composed of nine tests aimed at covering five different cognitive functions: attention (attentional control test), intellectual (thematic draws and conceptual activity test), memory (immediate memory and logical memory tests), oral language (receptive speech and expressive speech tests), and visuospatial (visual perception and spatial orientation tests) functions. Subjects under study performed the Luria-AND test at the beginning and at the end of the protocol. 2.4. Signal Pre-Processing A notch filter at 50 Hz was previously applied to the data to suppress power line noise. DC component and high frequencies were reduced with a band-pass finite impulse response (FIR) filter, employing a Hamming window in the frequency range 0.1–60 Hz. The filtered EEG was divided into epochs of 10 s each, without overlapping. In this way, visual inspection for artefacts could be carried out. After concluding the inspection of all recordings, no epochs were found that had to be rejected due to the presence of artefacts. This may be due to the fact that, during the 2 min EEG recording, users were asked to avoid eye movement or jaw clenching, which could introduce noise into the signal. 2.5. Metrics for Assessing EEG Changes In order to comprehensively assess EEG changes induced by an MI–NFT, both linear (relative power) and a non-linear (MSE) analyses were applied. 2.5.1. Relative Power The RP measurement gives information on the normalized weight of each band in the spectral distribution. The spectral bands selected to assess the induced EEG changes Entropy 2021,23, 1574 6 of 19 were delta (0.1–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), and beta (13–30 Hz). The lower ( l fi ) and upper ( u fi ) frequency limits of each band were specified according to the classical definitions of EEG bands [ 57 ]. The power spectral density (PSD) of the signal was calculated by means of the Welch method [ 58 ]. A Hamming window of 16 s (4096 samples; spectral resolution of 0.0625 Hz), along with a 90% of overlap and fast Fourier transform of 4096 points, was used. From this magnitude, the relative power (RP) of each band was estimated as follows: RPi=∑f=u fi f=l fiPSD(f) ∑f=60Hz f=0.1HzPSD(f),i={Delta,Theta,Alpha,Beta}(1) 2.5.2. Multiscale Entropy Traditional non-linear techniques, such as those based on entropy metrics, assess the irregularity of a time series in terms of the presence of repeating patterns [ 53 ]. Therefore, the highest values are assigned to random signals. Since complexity analysis methods should reflect the dynamic richness of a system, these measures must assign lowest values to both fully deterministic and fully random time series [ 53 ]. According to Costa et al. [ 53 ], biological systems need to operate across multiple spatial and temporal scales, which is also reflected in the complexity of their signals across scales. In this context, MSE is a measurement of complexity that fulfils the aforementioned requirement [ 47 , 59 ] and focuses on quantifying the information expressed by the physiologic dynamics over multiple time scales [53]. This is accomplished through estimation of the entropy (i.e., irregularity) on coarse-grained versions of the original signal. As a result of these calculations, MSE curves are obtained and can be used to compare the complexity of time series. The MSE curve whose entropy values are higher for the most of time scales is considered more complex [47,53]. In this study, we used SampEn as an irregularity metric. In this respect, irregularity is estimated as the negative logarithm of the conditional probability that two sequences of m (embedding dimension) consecutive data points, which fulfil a tolerance criterion (i.e., are considered similar), still meet the criterion when their lengths are increased in one sample (Equation (6)). Therefore, the higher the self-similarity in the time-series, the lower the estimation. Formally, given Ndata points from a time-series X = {x1,x2, . . . , xN} , these steps should be followed to estimate SampEn [60]: 1. FormN-m+1template vectorsofmconsecutivedatapoints {Xm(1), . . . , Xm(N−m+1)} , whereeachtemplatevectorisdefinedby Xm(i) = {xi, . . . , xi+m−1} ,i= {1, . . . , N−m+1} . These vectors represent mconsecutive values of Xcommencing with the i-th sample. 2. Define the distance between Xm(i) and Xm(j) , d[Xm(i) , Xm(j)] . In this context, Chebyshev is the most common metric used to compute the distance between vectors [59], d[Xm(i),Xm(j)] = maxk=1,...,m(|x(i+k−1)−x(j+k−1)|). (2) 3. Define a tolerance criterion in terms of the standard deviation ( σ ) of the time series, R=r·σ , where ris a parameter to set. Thus, two template vectors, Xm(i) and Xm(j) , are considered similar if their distance is less than the tolerance value: d[Xm(i),Xm(j)] <R. (3) 4. For each Xm(i) , count the number of vectors Xm(j) , given i 6= j, that fulfil the tolerance criterion (Equation (3)). This count is denoted as Bi . Then, the frequency of patterns similar to a given one of window length mis calculated by Bm i(r) = Bi N−m+1,i=1, . . . , N−m+1. (4) Entropy 2021,23, 1574 7 of 19 5. Define the probability that two sequences will match for mpoints as Bm(r) = 1 N−m N−m ∑ i=1 Bm i(r)(5) 6. Increase the embedding dimension to m+1, and repeat steps 1 to 5 in order to calculate Am+1(r) as the probability that two vectors, Xm+1(i) and Xm+1(j) , given i 6= j, match for m+1 points. 7. Then, we can estimate SampEn by computing: SampEn(m,r,N) = −lnAm+1(r) Bm(r). (6) SampEn is used to estimate the MSE because, unlike approximate entropy (ApEn) algorithm, SampEn does not count self-matching, which introduces an inherent bias, and the result provided is less dependent of time series length [ 53 ]. To obtain the τ -th timescaled version of the EEG, the original procedure was to divide the signal into N/ τ consecutive and non-overlapping segments of length τ and average the samples of each segment [ 47 ]. Nevertheless, these calculations may cause aliasing [ 59 ]. Therefore, as Martínez-Cagigal et al. [ 61 ] proposed, we estimated time-scaled versions by decimating the original signal. For the τ -th time scale, we applied a low-pass least-squares linearphase FIR filter to the pre-processed EEG signals ( ∼ 120 s duration) in order to reduce high frequencies. Then, we applied a downsampling procedure. That is, only every τ -th sample was kept. Finally, the irregularity of different time-scaled signals was estimated by the SampEn algorithm from original time series (i.e., τ = 1) up to the highest scale ( τ = 20), obtaining the MSE curves for each channel by plotting the results. In this regard, we considered the scales that fulfil the Richman & Moorman criterion (i.e., N≥10m) [60]. 2.6. Statistical Analysis The normality of the distributions of the pre- and post-training EEG measures, as well as the pre- and post-training Luria-AND scores, were explored by applying the Kolmogorov–Smirnov test. Neither EEG results (i.e., the spectral and complexity values) nor the neuropsychological scores fulfilled the parametric assumption. Thus, the non-parametric Wilcoxon signed-rank test was performed in order to assess the statistical differences between pre- and post-training EEG recordings, and analyse changes in cognitive functions. Additionally, the possible relationship between changes in RP of each frequency band and the change in EEG complexity after MI–NFT was analysed by Spearman’s rank correlation. In order to limit the number of comparisons, each MSE curve was characterised by a single value. This was obtained from the median value of the 20 time scales. We analysed each channel separately. Furthermore, to investigate the relationship between changes in MSE and neuropsychological improvements after MI–NFT, Spearman’s rank correlation was calculated separately for each Luria-AND test that showed a significant improvement. MSE curves were also characterised by estimating their median value. Finally, it is worth noting that, to overcome the problem of false discoveries due to multiple comparisons, the Benjamini–Hochberg false discovery rate (FDR–BH) correction was applied [62]. 3. Results 3.1. EEG Spectral Analysis Pre- and post-training PSDs were averaged across channels and subjects. As can be seen in Figure 3, a significant power increase ( p< 0.05) is found in theta, alpha, and beta frequencies, with a greater number of significant frequency bins in the alpha and beta bands. Differences between pre- and post-training RPs of each frequency band, averaged Entropy 2021,23, 1574 8 of 19 across channels and subjects, are displayed in Table 2. As shown, RP increased in theta ( p< 0.01), alpha ( p< 0.01), and beta ( p< 0.01) bands, while RP decreased in delta band (p<0.01), which means a power shift towards higher frequencies after NFT. Figure 3. The graph depicts the grand-average of the PSD across channels and subjects. Solid lines indicate the averaged PSD for pre- (red), and post-training (blue); whereas shaded areas indicate the 95% confidence interval. Frequency bands are indicated by dashed lines (delta: δ ; theta: θ ; alpha: α ; beta: β ). Wilcoxon signed-rank test p -values that show significant differences ( p< 0.05) between pre- and post-training PSD are shown in the bottom bar. FDR–BH correction was applied. Table 2. Pre- and post-training RP values of each frequency band, averaged across channels, and its standard deviation (std). The Wilcoxon signed-rank test p-values with FDR-BH correction are shown in the last column. Delta Theta Alpha Beta Pre-NFT RP (mean ±std) 0.622 ±0.154 0.060 ±0.029 0.112 ±0.068 0.132 ±0.073 Post-NFT RP (mean ±std) 0.520 ±0.131 0.073 ±0.028 0.149 ±0.077 0.175 ±0.080 p-value 0.0013 0.0013 0.0013 0.0034 Figure 4depicts the intra-channel comparison between pre- and post-training RP of each frequency band. Significant changes are found in all bands, especially in delta, alpha, and beta. Frontal channels (F3 and F4) show significant differences ( p< 0.01) in these three bands, whilst a significant increase ( p< 0.05) is also found at F4 in theta band. In addition, three other channels (C4, T8, and Pz) exhibit significant RP differences ( p< 0.05) in three out of the four frequency bands. It is noteworthy that all eight channels present significant RP differences (p<0.05) in at least one frequency band. Entropy 2021,23, 1574 9 of 19 Figure 4. Violin plots depict pre- (red, left side) and post-training (blue, right side) RP of delta ( A ), theta ( B ), alpha ( C ), and beta ( D ) frequency bands. Significant differences are marked with * ( p< 0.05) and ** ( p< 0.01). p-values were corrected with FDR–BH correction. 3.2. EEG Complexity Analysis To estimate the MSE curves, parameter values were varied in accordance with common ranges used for biomedical signals: embedding dimension m= {1,2} and tolerance parameter r= { 0.1, 0.15, 0.2, 0.25, 0.3 } [ 60 ]. Hereafter, we focus our discussion on the MSE curves obtained for m= 2 and r= 0.1 (i.e., R= 0.1 ·σ ), since these values are widely used in the literature [ 60 , 63 ]. However, this selection of parameters did not influence the results obtained, showing their consistency across the different values. MSE curves calculated for the remaining parameter values can be found in the supplementary material. MSE curves for each channel are displayed in Figure 5. Differences were evaluated for each time scale in each channel. A general increase in SampEn values compared to the pre-training ones is observed, except for Cz. Significant increases are found in five of the eight channels: C4 and Pz ( p< 0.05), and F3, F4, and T8 ( p< 0.01). In this regard, it is remarkable that the F3 and F4 channels show significant differences for all time scales considered, whilst significant differences are found from τ= 3 onwards in C4, and from τ= 5 onwards in T8. On the other hand, Pz curves show significant differences for all time scales, except for τ={1, 2,8}. Entropy 2021,23, 1574 16 of 19 and may be helpful to reach a deeper understanding of how neurofeedback affects the brain after training. Furthermore, the analysis of scores from neuropsychological tests showed an improvement in intellectual and memory functions in elderly people after MI- based NFT. Accordingly, MI–NFT may help the elderly counteract the effects of age-related cognitive decline. Supplementary Materials: The following are available online at https://www.mdpi.com/1099-430 0/23/12/1574/s1, Figure S1: MSE values across channels (m= 1; r= 0.1). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S2: MSE values across channels (m= 1; r= 0.15). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S3: MSE values across channels (m= 1; r= 0.2). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S4: MSE values across channels (m= 1; r= 0.25). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S5: MSE values across channels (m= 1; r= 0.3). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S6: MSE values across channels (m= 2; r= 0.1). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S7: MSE values across channels (m= 2; r= 0.15). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S8: MSE values across channels (m= 2; r= 0.2). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S9: MSE values across channels (m= 2; r= 0.25). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Figure S10: MSE values across channels (m= 2; r= 0.3). Lines with circle markers indicate pre- (red) and post-training (grey) averaged sample entropy values. Significant differences ( p< 0.05) for each time scale are shown in the bottom bars. P-values were corrected with FDR-BH correction. Author Contributions: Investigation, D.M.-M.; Methodology, D.M.-M., V.M.-C., E.S.-V. and R.H.; Resources, V.M.-C. and R.H.; Supervision, V.M.-C., E.S.-V. and R.H.; Writing—original draft, D.M.-M.; Writing—review and editing, D.M.-M., V.M.-C., E.S.-V., S.P.-V. and R.H. All authors have read and agreed to the published version of the manuscript. Funding: This research has been developed under the grants PID2020-115468RB-I00 and RTC2019- 007350-1 funded by ‘Ministerio de Ciencia e Innovación/Agencia Estatal de Investigación/10.13039/ 501100011033/’ and ERDF A way of making Europe; under the R+D+i project ‘Análisis y correlación entre la epigenética y la actividad cerebral para evaluar el riesgo de migraña crónica y episódica en mujeres’ (‘Cooperation Programme Interreg V-A Spain-Portugal POCTEP 2014–2020’) funded by ‘European Commission’ and ERDF; and by ‘CIBER en Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN)’ through ‘Instituto de Salud Carlos III’ co-funded with ERDF funds. Diego Marcos- Martínez, Eduardo Santamaría-Vázquez and Sergio Pérez-Velasco were in receipt of a PIF grant by the ’Consejería de Educación de la Junta de Castilla y León’. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study or their caregivers. Data Availability Statement: The dataset analysed in the study is available from the authors upon reasonable request. Conflicts of Interest: The authors declare no conflict of interest. Entropy 2021,23, 1574 17 of 19 References 1. Wolpaw, J.R.; Birbaumer, N.; McFarland, D.J.; Pfurtscheller, G.; Vaughan, T.M. Brain-computer interfaces for communication and control. Clin. Neurophysiol. 2002,113, 767–791. [CrossRef] 2. Wolpaw, J.; Wolpaw, E.W. Brain-Computer Interfaces: Principles and Practice; Oxford University Press: Oxford, UK, 2012. 3. Mane, R.; Chouhan, T.; Guan, C. BCI for stroke rehabilitation: Motor and beyond. J. Neural Eng. 2020,17, 041001. [CrossRef] 4. Sebastián-Romagosa, M.; Cho, W.; Ortner, R.; Murovec, N.; Von Oertzen, T.; Kamada, K.; Allison, B.Z.; Guger, C. Brain Computer Interface Treatment for Motor Rehabilitation of Upper Extremity of Stroke Patients—A Feasibility Study. Front. Neurosci. 2020 , 14, 1056. [CrossRef] 5. Enriquez-Geppert, S.; Huster, R.J.; Herrmann, C.S. EEG-neurofeedback as a tool to modulate cognition and behavior: A review tutorial. Front. Hum. Neurosci. 2017,11, 51. [CrossRef] [PubMed] 6. Sitaram, R.; Ros, T.; Stoeckel, L.; Haller, S.; Scharnowski, F.; Lewis-Peacock, J.; Weiskopf, N.; Blefari, M.L.; Rana, M.; Oblak, E.; Birbaumer, N.; Sulzer, J. Closed-loop brain training: The science of neurofeedback. Nat. Rev. Neurosci. 2017 ,18, 86–100. [CrossRef] [PubMed] 7. Ros, T.; Munneke, M.A.; Ruge, D.; Gruzelier, J.H.; Rothwell, J.C. Endogenous control of waking brain rhythms induces neuroplasticity in humans. Eur. J. Neurosci. 2010,31, 770–778. [CrossRef] [PubMed] 8. Arns, M.; Heinrich, H.; Strehl, U. Evaluation of neurofeedback in ADHD: The long and winding road. Biol. Psychol. 2014 , 95, 108–115. [CrossRef] [PubMed] 9. Arns, M.; Clark, C.R.; Trullinger, M.; DeBeus, R.; Mack, M.; Aniftos, M. Neurofeedback and Attention-Deficit/Hyperactivity- Disorder (ADHD) in Children: Rating the Evidence and Proposed Guidelines. Appl. Psychophysiol. Biofeedback 2020 ,45, 39–48. [CrossRef] 10. Tan, G.; Thornby, J.; Hammond, D.C.; Strehl, U.; Canady, B.; Arnemann, K.; Kaiser, D.A. Meta-analysis of EEG biofeedback in treating epilepsy. Clin. EEG Neurosci. 2009,40, 173–179. [CrossRef] 11. Prichep, L.S.; John, E.R.; Ferris, S.H.; Reisberg, B.; Almas, M.; Alper, K.; Cancro, R. Quantitative EEG correlates of cognitive deterioration in the elderly. Neurobiol. Aging 1994,15, 85–90. [CrossRef] 12. Jeong, J. EEG dynamics in patients with Alzheimer’s disease. Clin. Neurophysiol. 2004,115, 1490–1505. [CrossRef] [PubMed] 13. Czigler, B.; Csikós, D.; Hidasi, Z.; Anna Gaál, Z.; Csibri, É.; Kiss, É.; Salacz, P.; Molnár, M. Quantitative EEG in early Alzheimer’s disease patients—Power spectrum and complexity features. Int. J. Psychophysiol. 2008,68, 75–80. [CrossRef] 14. Benwell, C.S.; Davila-Pérez, P.; Fried, P.J.; Jones, R.N.; Travison, T.G.; Santarnecchi, E.; Pascual-Leone, A.; Shafi, M.M. EEG spectral power abnormalities and their relationship with cognitive dysfunction in patients with Alzheimer’s disease and type 2 diabetes. Neurobiol. Aging 2020,85, 83–95. [CrossRef] 15. Laborda-Sánchez, F.; Cansino, S. The Effects of Neurofeedback on Aging-Associated Cognitive Decline: A Systematic Review. Appl. Psychophysiol. Biofeedback 2021,46, 1–10. [CrossRef] [PubMed] 16. United Nations. World Population Ageing; Department of Economic and Social Affairs: New York, NY, USA, 2020; pp. 40–1307. 17. Wang, J.R.; Hsieh, S. Neurofeedback training improves attention and working memory performance. Clin. Neurophysiol. 2013 , 124, 2406–2420. [CrossRef] 18. Enriquez-Geppert, S.; Huster, R.J.; Scharfenort, R.; Mokom, Z.N.; Zimmermann, J.; Herrmann, C.S. Modulation of frontal-midline theta by neurofeedback. Biol. Psychol. 2014,95, 59–69. [CrossRef] 19. Vernon, D.; Egner, T.; Cooper, N.; Compton, T.; Neilands, C.; Sheri, A.; Gruzelier, J. The effect of trainin distinct neurofeedback protocols on aspects of cognitive performance. Int. J. Psuchophysiol. 2003,47, 75–85. [CrossRef] 20. Zoefel, B.; Huster, R.J.; Herrmann, C.S. Neurofeedback training of the upper alpha frequency band in EEG improves cognitive performance. NeuroImage 2011,54, 1427–1431. [CrossRef] [PubMed] 21. Kober, S.E.; Witte, M.; Stangl, M.; Väljamäe, A.; Neuper, C.; Wood, G. Shutting down sensorimotor interference unblocks the networks for stimulus processing: An SMR neurofeedback training study. Clin. Neurophysiol. 2015,126, 82–95. [CrossRef] 22. Campos da Paz, V.K.; Garcia, A.; Campos da Paz Neto, A.; Tomaz, C. SMR neurofeedback training facilitates working memory performance in healthy older adults: A behavioral and EEG study. Front. Behav. Neurosci. 2018,12, 321. [CrossRef] 23. Marlats, F.; Bao, G.; Chevallier, S.; Boubaya, M.; Djabelkhir-Jemmi, L.; Wu, Y.H.; Lenoir, H.; Rigaud, A.S.; Azabou, E. SMR/Theta Neurofeedback Training Improves Cognitive Performance and EEG Activity in Elderly With Mild Cognitive Impairment: A Pilot Study. Front. Aging Neurosci. 2020,12, 147. [CrossRef] 24. Becerra, J.; Fernández, T.; Roca-Stappung, M.; Díaz-Comas, L.; Galán, L.; Bosch, J.; Espino, M.; Moreno, A.J.; Harmony, T. Neurofeedback in healthy elderly human subjects with electroencephalographic risk for cognitive disorder. J. Alzheimer’s Dis. 2012,28, 357–367. [CrossRef] 25. Staufenbiel, S.M.; Brouwer, A.M.; Keizer, A.W.; van Wouwe, N.C. Effect of beta and gamma neurofeedback on memory and intelligence in the elderly. Biol. Psychol. 2014,95, 74–85. [CrossRef] 26. Luijmes, R.E.; Pouwels, S.; Boonman, J. L’efficacité du neurofeedback sur le fonctionnement cognitif chez les patients atteints de la maladie d’Alzheimer: Résultats préliminaires. Neurophysiol. Clin. 2016,46, 179–187. [CrossRef] 27. Pfurtscheller, G. Event-related synchronization (ERS): An electrophysiological correlate of cortical areas at rest. Electroencephalogr. Clin. Neurophysiol. 1992,83, 62–69. [CrossRef] 28. Sterman, M.B. Physiological origins and functional correlates of EEG rhythmic activities: Implications for self-regulation. Biofeedback Self-Regul. 1996,21, 3–33. [CrossRef] Entropy 2021,23, 1574 18 of 19 29. Neuper, C.; Scherer, R.; Wriessnegger, S.; Pfurtscheller, G. Motor imagery and action observation: Modulation of sensorimotor brain rhythms during mental control of a brain–computer interface. Clin. Neurophysiol. 2009 ,120, 239–247. [CrossRef] [PubMed] 30. Gomez-Pilar, J.; Corralejo, R.; Nicolas-Alonso, L.F.; Álvarez, D.; Hornero, R. Neurofeedback training with a motor imagery-based BCI: Neurocognitive improvements and EEG changes in the elderly. Med. Biol. Eng. Comput. 2016,54, 1655–1666. [CrossRef] 31. Mulder, T. Motor imagery and action observation: Cognitive tools for rehabilitation. J. Neural Transm. 2007 ,114, 1265–1278. [CrossRef] [PubMed] 32. Vourvopoulos, A.; Jorge, C.; Abreu, R.; Figueiredo, P.; Fernandes, J.C.; Bermúdez i Badia, S. Efficacy and brain imaging correlates of an immersive motor imagery BCI-driven VR system for upper limb motor rehabilitation: A clinical case report. Front. Hum. Neurosci. 2019,13, 244. [CrossRef] [PubMed] 33. Pichiorri, F.; Mattia, D. Brain-Computer Interfaces in Neurologic Rehabilitation Practice, 1st ed.; Elsevier B.V.: Amsterdam, The Netherlands, 2020; Volume 168, pp. 101–116. [CrossRef] 34. Yang, S.; Li, R.; Li, H.; Xu, K.; Shi, Y.; Wang, Q.; Yang, T.; Sun, X. Exploring the Use of Brain-Computer Interfaces in Stroke Neurorehabilitation. BioMed Res. Int. 2021,2021, 9967348. [CrossRef] [PubMed] 35. Baniqued, P.D.E.; Stanyer, E.C.; Awais, M.; Alazmani, A.; Jackson, A.E.; Mon-Williams, M.A.; Mushtaq, F.; Holt, R.J. Braincomputer interface robotics for hand rehabilitation after stroke: A systematic review. J. NeuroEng. Rehabil. 2021 ,18, 15. [CrossRef] 36. Cervera, M.A.; Soekadar, S.R.; Ushiba, J.; Millán, J.d.R.; Liu, M.; Birbaumer, N.; Garipelli, G. Brain-computer interfaces for post-stroke motor rehabilitation: A meta-analysis. Ann. Clin. Transl. Neurol. 2018,5, 651–663. [CrossRef] [PubMed] 37. De Castro-Cros, M.; Sebastian-Romagosa, M.; Rodríguez-Serrano, J.; Opisso, E.; Ochoa, M.; Ortner, R.; Guger, C.; Tost, D. Effects of Gamification in BCI Functional Rehabilitation. Front. Neurosci. 2020,14, 882. [CrossRef] 38. Klimesch, W. EEG alpha and theta oscillations reflect cognitive and memory performance: A review and analysis. Brain Res. Rev. 1999,29, 169–195. [CrossRef] 39. Bastiaansen, M.; Hagoort, P. Chapter 12 Oscillatory neuronal dynamics during language comprehension. Prog. Brain Res. 2006 , 159, 179–196. [CrossRef] 40. Yılmaz, N.H.; Çalı¸so˘glu, P.; Güntekin, B.; Hano˘glu, L. Correlation between alpha activity and neuropsychometric tests in Parkinson’s disease. Neurosci. Lett. 2020,738. [CrossRef] 41. Doppelmayr, M.; Klimesch, W.; Stadler, W.; Pöllhuber, D.; Heine, C. EEG alpha power and intelligence. Intelligence 2002 , 30, 289–302. [CrossRef] 42. Klimesch, W. Alpha-band oscillations, attention, and controlled access to stored information. Trends Cogn. Sci. 2012 ,16, 606–617. [CrossRef] [PubMed] 43. Lundqvist, M.; Herman, P.; Warden, M.R.; Brincat, S.L.; Miller, E.K. Gamma and beta bursts during working memory readout suggest roles in its volitional control. Nat. Commun. 2018,9, 394. [CrossRef] 44. Schmidt, R.; Ruiz, M.H.; Kilavik, B.E.; Lundqvist, M.; Starr, P.A.; Aron, A.R. Beta oscillations in working memory, executive control of movement and thought, and sensorimotor function. J. Neurosci. 2019,39, 8231–8238. [CrossRef] 45. Weiss, S.; Mueller, H.M. “Too many betas do not spoil the broth”: The role of beta brain oscillations in language processing. Front. Psychol. 2012,3, 201. [CrossRef] [PubMed] 46. Egner, T.; Gruzelier, J.H. EEG Biofeedback of low beta band components: Frequency-specific effects on variables of attention and event-related brain potentials. Clin. Neurophysiol. 2004,115, 131–139. [CrossRef] 47. Costa, M.; Goldberger, A.L.; Peng, C.K. Multiscale Entropy Analysis of Complex Physiologic Time Series. Phys. Rev. Lett. 2002 , 89, 6–9. [CrossRef] 48. Escudero, J.; Abásolo, D.; Hornero, R.; Espino, P.; López, M. Analysis of electroencephalograms in Alzheimer’s disease patients with multiscale entropy. Physiol. Meas. 2006,27, 1091. [CrossRef] 49. Hornero, R.; Abásolo, D.; Escudero, J.; Gómez, C. Nonlinear analysis of electroencephalogram and magnetoencephalogram recordings in patients with Alzheimer ’ s disease Receive free email alerts when new articles cite this article-sign up. Philos. Trans. R. Soc. A 2009,367, 317–336. [CrossRef] 50. Takahashi, T.; Cho, R.Y.; Murata, T.; Mizuno, T.; Kikuchi, M.; Mizukami, K.; Kosaka, H.; Takahashi, K.; Wada, Y. Age-related variation in EEG complexity to photic stimulation: A multiscale entropy analysis. Clin. Neurophysiol. 2009 ,120, 476–483. [CrossRef] 51. Maturana-Candelas, A.; Gómez, C.; Poza, J.; Pinto, N.; Hornero, R. EEG characterization of the Alzheimer’s disease continuum by means of multiscale entropies. Entropy 2019,21, 544. [CrossRef] [PubMed] 52. Su, R.; Li, X.; Liu, Y.; Cui, W.; Xie, P.; Han, Y. Evaluation of the Brain Function State During Mild Cognitive Impairment Based on Weighted Multiple Multiscale Entropy. Front. Aging Neurosci. 2021,13, 430. [CrossRef] 53. Costa, M.; Goldberger, A.L.; Peng, C.K. Multiscale entropy analysis of biological signals. Phys. Rev. E Stat. Nonlinear Soft Matter Phys. 2005,71, 021906. [CrossRef] [PubMed] 54. Christensen, A.L. A Practical Application of the Luria Methodology. J. Clin. Neuropsychol. 1979 ,1, 241–247. 8414453. [CrossRef] 55. Klem, G.H.; Lüders, H.O.; Henri, H.; Elger, C. The ten ± twenty electrode system of the International Federation. Electroencephalogr. Clin. Neurophysiol. 1958,10, 371–375. 56. Schalk, G.; Mcfarland, D.J.; Hinterberger, T.; Birbaumer, N.; Wolpaw, J.R.; Technology 753 ABcIBCI. BCI2000: A General-Purpose Brain-Computer Interface (BCI) 754 System. IEEE Trans. Biomed. Eng. 2004,51, 1034–1043. [CrossRef] [PubMed] Entropy 2021,23, 1574 19 of 19 57. Kane, N.; Acharya, J.; Benickzy, S.; Caboclo, L.; Finnigan, S.; Kaplan, P.W.; Shibasaki, H.; Pressler, R.; van Putten, M.J. A revised glossary of terms most commonly used by clinical electroencephalographers and updated proposal for the report format of the EEG findings. Revision 2017. Clin. Neurophysiol. Pract. 2017,2, 170–185. [CrossRef] 58. Welch, P.D. The Use of Fast Fourier Transform for the Estimation of Power Spectra. IEEE Trans. Audio Electroacoust. 1976 ,15, 70–73. [CrossRef] 59. Humeau-Heurtier, A. The multiscale entropy algorithm and its variants: A review. Entropy 2015,17, 3110–3123. [CrossRef] 60. Richman, J.S.; Moorman, J.R. Physiological time-series analysis using approximate and sample entropy. Am. J. Physiol. Heart Circ. Physiol. 2000,278, H2039–H2049. [CrossRef] [PubMed] 61. Martínez-Cagigal, V.; Santamaría-Vázquez, E.; Hornero, R. Asynchronous control of P300-based brain–computer interfaces using sample entropy. Entropy 2019,21, 230. [CrossRef] 62. Benjamini, Y.; Hochberg, Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J. R. Stat. Soc. Ser. B Methodol. 1995,57, 289–300. [CrossRef] 63. Amarantidis, L.C.; Abásolo, D. Interpretation of entropy algorithms in the context of biomedical signal analysis and their application to EEG analysis in epilepsy. Entropy 2019,21, 840. [CrossRef] 64. Duncan, J.; Emslie, H.; Williams, P.; Johnson, R.; Freer, C. Intelligence and the frontal lobe: The organization of goal-directed behavior. Cogn. Psychol. 1996,30, 257–303. [CrossRef] 65. Duncan, J.; Seitz, R.J.; Kolodny, J.; Bor, D.; Herzog, H.; Ahmed, A.; Newell, F.N.; Emslie, H. A neural basis for general intelligence. Science 2000,289, 457–460. [CrossRef] [PubMed] 66. Fuster, J.M. Frontal lobe and cognitive development. J. Neurocytol. 2002,31, 373–385. [CrossRef] [PubMed] 67. Elhalal, A.; Davelaar, E.J.; Usher, M. The role of the frontal cortex in memory: An investigation of the Von Restorff effect. Front. Hum. Neurosci. 2014,8, 410. [CrossRef] [PubMed] 68. Cavanagh, J.F.; Frank, M.J. Frontal theta as a mechanism for cognitive control. Trends Cogn. Sci. 2014,18, 414–421. [CrossRef] 69. Fernández, A.; Pinal, D.; Díaz, F.; Zurrón, M. Working memory load modulates oscillatory activity and the distribution of fast frequencies across frontal theta phase during working memory maintenance. Neurobiol. Learn. Mem. 2021 ,183, 107476. [CrossRef] 70. Jurewicz, K.; Paluch, K.; Kublik, E.; Rogala, J.; Mikicin, M.; Wróbel, A. EEG-neurofeedback training of beta band (12–22 Hz) affects alpha and beta frequencies—A controlled study of a healthy population. Neuropsychologia 2018,108, 13–24. [CrossRef] 71. De Munck, J.C.; Gonçalves, S.I.; Mammoliti, R.; Heethaar, R.M.; Lopes da Silva, F.H. Interactions between different EEG frequency bands and their effect on alpha-fMRI correlations. NeuroImage 2009,47, 69–76. [CrossRef] 72. Mizuno, T.; Takahashi, T.; Cho, R.Y.; Kikuchi, M.; Murata, T.; Takahashi, K.; Wada, Y. Assessment of EEG dynamical complexity in Alzheimer’s disease using multiscale entropy. Clin. Neurophysiol. 2010,121, 1438–1446. [CrossRef] 73. Goldberger, A.L.; Peng, C.K.; Lipsitz, L.A. What is physiologic complexity and how does it change with aging and disease? Neurobiol. Aging 2002,23, 23–26. [CrossRef] 74. Van Noordt, S.; Willoughby, T. Cortical maturation from childhood to adolescence is reflected in resting state EEG signal complexity. Dev. Cogn. Neurosci. 2021,48, 100945. [CrossRef] [PubMed] 75. Ero˘glu, G.; Gürkan, M.; Teber, S.; Ertürk, K.; Kırmızı, M.; Ekici, B.; Arman, F.; Balcisoy, S.; Özgüz, V.; Çetin, M. Changes in EEG complexity with neurofeedback and multi-sensory learning in children with dyslexia: A multiscale entropy analysis. Appl. Neuropsychol. Child 2020,9, 1–12. [CrossRef] 76. Angelakis, E.; Stathopoulou, S.; Frymiare, J.L.; Green, D.L.; Lubar, J.F.; Kounios, J. EEG neurofeedback: A brief overview and an example of peak alpha frequency training for cognitive enhancement in the elderly. Clin. Neuropsychol. 2007 ,21, 110–129. [CrossRef] [PubMed] 77. Ros, T.; Enriquez-Geppert, S.; Zotev, V.; Young, K.D.; Wood, G.; Whitfield-Gabrieli, S.; Wan, F.; Vuilleumier, P.; Vialatte, F.; Van De Ville, D.; et al. Consensus on the reporting and experimental design of clinical and cognitive-behavioural neurofeedback studies (CRED-nf checklist). Brain 2020,143, 1674–1685. [CrossRef] [PubMed]