FBG sensor for heart rate monitoring using 3D printing technology

Received 12 March 2024, accepted 10 April 2024, date of publication 15 April 2024, date of current version 26 April 2024. Digital Object Identifier 10.1109/ACCESS.2024.3389495 FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology M. FAJKUS 1, M. KOSTELANSKY 1, M. FRIDRICH 1, J. CUBIK 1, S. KEPAK 1, D. KRIZAN 1, R. MARTINEK 2, (Senior Member, IEEE), MAZIN ABED MOHAMMED 1,2,3, AND J. NEDOMA 1, (Senior Member, IEEE) 1Department of Telecommunications, Faculty of Electrical Engineering and Computer Science, VSB—Technical University of Ostrava, 70800 Ostrava, Czech Republic 2Department of Cybernetics and Biomedical Engineering, Faculty of Electrical Engineering and Computer Science, VSB—Technical University of Ostrava, 70800 Ostrava, Czech Republic 3Department of Artificial Intelligence, College of Computer Science and Information Technology, University of Anbar, Anbar 31001, Iraq Corresponding author: M. Fajkus ([email protected]) This work was supported in part by the European Union through the REFRESH—Research Excellence For REgion Sustainability and High-tech Industries under Project CZ.10.03.01/00/22_003/0000048 via the Operational Program Just Transition; and in part by the Ministry of Education, Youth and Sports of the Czech Republic conducted by the VSB—Technical University of Ostrava under Grant SP2024/081. This work involved human subjects or animals in its research. Approval of all ethical and experimental procedures and protocols was granted by the Ethics Commission at VSB-TU Ostrava under Application No. VSB/23/109812. ABSTRACT Currently, the use of fiber-optic Bragg gratings in biomedical applications, especially in the field of magnetic resonance imaging (MRI), is becoming popular. In these applications, the fiber Bragg grating (FBG) encapsulation plays a crucial role in terms of the accuracy and reproducibility of the measurements. This paper describes in detail the fabrication method of a prototype FBG sensor, which is realized by encapsulating a Bragg grating between two layers of the MR-compatible material Acrylonitrile Butadiene Styrene (ABS) by 3D printing. The sensor thus created, implemented, for example, on the chest of a human body, enables monitoring of the vital functions of the human body. The paper describes the complete procedure for the creation of the prototype sensor, including strain and temperature dependence, as well as results of long-term experimental measurements against the conventional electrocardiography (ECG) standard. Results based on the objective Bland-Altman (B-A) method confirm that the implemented sensor can be used for reliable monitoring of cardiac activity (>95% based on B-A). Taking into account the single fiber optic cable, its simple implementation, its small size and weight < 5g, the presented sensor represents an interesting alternative to conventional ECG. INDEX TERMS Fibre Bragg grating, 3D printing, vital sign monitoring, heart rate, MR-compatible. I. INTRODUCTION Monitoring cardiac activity is a key element in the diagnosis and tracking of the health status of patients. Heart rate and its variability provide valuable information on heart function and the general condition of the cardiovascular system. Electrocardiography (ECG) is a standard and widely used method to record electrical activity of the heart, which The associate editor coordinating the review of this manuscript and approving it for publication was Salvatore Surdo . allows the identification of potential arrhythmias, ischemia, or other abnormalities. ECG records the electrical activity of the heart, resulting in an electrocardiogram containing a sequence of PQRST waves [1],[2]. A very specific environment is magnetic resonance imaging, which uses strong magnetic fields to create detailed images of the internal organs and structures of the human body. In some cases, cardiac activity monitoring is used during magnetic resonance imaging. These cases include the investigation of patients with heart disease or high-risk conditions 57150 2024 The Authors. This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ VOLUME 12, 2024 M. Fajkus et al.: FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology (e.g., a patient with a pacemaker), but also in the case of imaging the heart itself, where it is necessary to synchronize the imaging with the activity of the heart to achieve sharp and clear images. The standard method for monitoring cardiac activity in an MRI environment is an MR-compatible ECG [3]. However, these lead to the problem of the magnetohydrodynamic principle of electrically conducting blood, whose electric current is added to the cardiac signal, increasing the t-wave amplitude in the MR environment. This overshoot of the QRS complex causes inaccurate synchronization, leading to less sharp images and prolonged examinations [4]. Alternatives to MR-compatible ECG are methods such as photoplethysmography [5] or acoustic systems [6], which are fully immune to electromagnetic fields. In addition to monitoring cardiac activity in the MR environment, respiratory monitoring is also performed, where the tunnel environment can cause hyperventilation due to the shortness of breath (dyspnea) [7]. The above-mentioned conventional approaches are used to monitor fundamental parameters of the human body in long-term or intensive care units, sleep laboratories, or MRI environments. An alternative approach is fiber optic technology,whichis immuneto electromagneticinterference, has a very small footprint, and can be operated with a single lead optical cable. The first group is represented by intensity (micro bending) sensors based on the evaluation of changes in the optical power of light passing through optical fibers [8]. Usually, these sensors are placed in the pad on which the patient is lying. During respiration, there is a variable load on the sensor that results in modulation of intensity of the transmitted or reflected light [9]. Another group of fiber-optic sensors uses the principle of light interference. These sensors exhibit high sensitivity in measuring vibration, acceleration, and derived quantities and are found in Mach-Zehnder [10], Michelson [11], Sagnac [12] or Fabry-Perot [13] interferometer arrangements. The Mach-Zehnder-based interferometric approach represents the simplest option for monitoring vital signs of the human body [10]. Due to its high sensitivity, it can record both respiratory and cardiac activity [14]. The third group consists of sensors that use fiber Bragg gratings (FBG) [15], which are the fastest growing fiber optic sensor technology in biomedical applications [16]. These sensors are also sensitive enough to monitor cardiac activity in addition to respiratory activity [17]. The use of FBG sensors in monitoring of respiratory and cardiac activity requires appropriate encapsulation of the Bragg grating, which aims to achieve the required sensitivity to the pressure caused by respiration and cardiac activity and to provide protection against damage. The use of FBG sensors to monitor respiratory and cardiac activity requires appropriate encapsulation of the Bragg grating, which aims to achieve the required sensitivity to pressure caused by respiration and cardiac activity and to provide protection against damage. The simplest encapsulation methods include the attachment of an optical fiber with a Bragg grating to a flexible plastic pad on which the subject is lying [17]. Increasing the precision of the respiration and heart rate can be achieved by implementing a cascade connection of several Bragg gratings with a single fiber optic cable [18]. Another option is to implement an FBG sensor on the chest in the region of the heart. These sensors consist of a Bragg grating encapsulated typically in polymers (e.g., polydimethylsiloxane PDMS) and fixed with an elastic chest band [19],[20]. Due to their location close to the heart, they exhibit high sensitivity to cardiac activity [21],[22]. An interesting approach involves encapsulating FBG sensors within smart patches for the simultaneous monitoring of respiratory and heart activity. The study demonstrates the high fidelity of the smart patch in estimating heart rate (HR) and respiratory rate (RR) under various respiratory conditions and common daily body positions, while maintaining patient comfort throughout the examination [23]. The most recent approach is to encapsulate Bragg gratings in mattresses, beds [24] and smart textiles [25]. The emerging field of Bragg grating encapsulation uses 3D printing technology. The authors of papers [26],[27],[28], [29],[30],[31] presented experiments on the encapsulation of FBGs in polymer structures. Encapsulation using 3D printing technology was found to not affect the functionality of the FBG or the shape of the reflectance spectrum [30], only the reflectance spectrum is shifted and the sensitivity to mechanical loading is affected. For example, when embedded in a thermoplastic polyurethane (TPU) material, a shift of the reflectance spectrum by 136 pm was observed and also a bending sensitivity of about 1.74×compared to bare fiber with a Bragg grating has been demonstrated [31]. However, the encapsulation material itself has a significant effect on the temperature sensitivity, which can increase more than tenfold [27],[28],[32]. The authors of the paper [33] presented the application of encapsulated FBG sensors using 3D printing technology to monitor structural stresses, presenting a prototype with three Bragg gratings for effective stress change observation. Further analysis revealed that FBG sensors with polyimide-coated fibers encapsulated in Acrylonitrile Butadiene Styrene (ABS) exhibited significantly higher strain sensitivity, approximately 2.5 times greater, compared to those with acrylate protection and Polylactic Acid (PLA) encapsulation. [34] A more detailed analysis compared, in addition to the above materials, Ormocer primary protection and TPU filament [35]. The temperature loading results show that ABS samples exhibit a temperature sensitivity of approximately 116 pm/◦C in the range up to 60 ◦C, while PLA samples exhibit a temperature sensitivity of up to 139 pm/◦C, but cannot be used at temperatures higher than 40 ◦C. In the case of tensile tests, the highest response was achieved for samples made of TPU filament. VOLUME 12, 2024 57151 M. Fajkus et al.: FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology Other applications of sensors based on Bragg gratings encapsulated using 3D printing technology that have been published include pressure sensors [26], tilt sensors with a sensitivity of 10 pm/◦[36] or, for example, special clamps for mounting an industrial hydraulic piping system to detect its loosening or determine its natural resonant frequency [37]. Although various studies have explored the use of 3D printing and fiber optic sensors for diverse applications, this publication uniquely focuses on utilizing 3D-printed FBG sensors for vital sign monitoring. Previous works [38] and [39] have explored the measurement of heart and breathing rates with 3D-printed encapsulated Bragg gratings, while [40] employs PLA filament for FBG sensor encapsulation, unlike using the ABS filament described in this article, which has different mechanical properties that affect sensor performance. Additionally, such FBG sensors have been applied in biomedical devices such as spirometers [41] and respiratory activity sensors embedded in oxygen masks [42], offering advantages in MRI environments by reducing motion artifacts. In [43], a thumb-shaped FBG sensor encapsulated with Fused Deposition Modeling (FDM) technology was developed for use under an anesthesiologist’s glove to precisely locate the epidural space. The optical signal changes when the needle contacts the epidural space, helping anesthesiologists in real-time needle positioning to improve procedure success rates. Table 1compares the method of encapsulating Bragg gratings using 3D printing technology and its effect on temperature and strain sensitivity as a function of filament type. The results show that, for example, the resulting temperature sensitivity is not only influenced by the filament type itself, but also by the specific filament composition or by the dimensions and method of imprinting. TABLE 1. Comparison of strain and temperature sensitivities of FBGs encapsulated using 3D printing technology. The above review shows a number of advantages of fiber optic Bragg gratings in combination with 3D printing encapsulation technology even in biomedical applications, specifically in the MRI environment. The authors of this publication see the benefit in the FBG encapsulation detailed description of the method using the 3D printing method, which results in a sensor with very small dimensions of 10 ×5×2 mm and a weight of less than 5 g. Furthermore, the temperature and deformation characteristics of the sensor and results from long-term measurements of heart activity (heart rate) are described, including different ways to fix the sensor on the human body, all compared to a conventional ECG. The encapsulation process itself offers easy reproducibility and the sensor can be used repeatedly. Based on the measurements made, the sensor prototype represents an interesting alternative solution for the needs of sleep laboratories, magnetic resonance imaging environments, or environments in general where it is necessary to monitor cardiac activity over time. II. METHODS A. FIBER BRAGG GRATING (FBG) The FBG is formed by a periodic change of refractive index in the core of an optical fiber. This periodic structure reflects a narrow part of the spectrum and transmits the rest of the spectrum without attenuation. The reflected wavelength is called the Bragg wavelength and is given by equation: λB=2neff 3, (1) where neff is the effective refractive index of the grating structure and 3is the period of refractive index changes. If temperature or strain acts on this structure, periodic changes occur, and the effective refractive index changes. This causes a shift of the reflected Bragg wavelength. The dependence of the Bragg wavelength on the applied temperature and strain is expressed by the following equation: 1λB λB =(1−pe)ε+(αn+α3)1T,(2) where peis the photo-elastic coefficient, αnis the temperature-optic coefficient and α3is the thermal expansion coefficient. The temperature and strain coefficient vary with the absolute value of the Bragg wavelength. Therefore, a normalized strain coefficient at constant temperature is: 1 λB 1λB 1ε =0.78 ·10−6µstrain−1(3) and a normalized temperature coefficient at constant strain is: 1 λB 1λB 1T=6.678 ·10−6◦C−1.(4) B. BALLISTOCARDIOGRAPHY Using the above-mentioned FBG sensor, the so-called ballistocardiograph signal (BCG) can be obtained. This method falls into the category of non-invasive sensing of body movements caused by the heart. Body movements measured by this method are caused by the acceleration of blood as it moves within large blood vessels [44]. Specifically, the consequence of blood impingement on the so-called aortic arch, which causes upward movement of the body and subsequent downward movement of the body as blood descends, is sensed. The current trend of using BCG type measurements is starting to increase again, thanks to new technologies, especially in the field of fiber-optic sensors; see Introduction. The progression or comparison of ECG and BCG is shown in Figure 1. 57152 VOLUME 12, 2024 M. Fajkus et al.: FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology FIGURE 1. Course and comparison of ECG and BCG signals. III. RESULTS A. METHOD OF MEASURING HEART RATE WITH FBG Due to the need to monitor cardiac activity, the position chosen for the experimental measurements was in the chest area, on the left side in the region of the heart (Figure 2marked in red). This position was chosen based on laboratory measurements primarily because it allows the contact area of the sensor with the human body, regardless of gender; see our previous publications where this issue was addressed [45]. To achieve the highest sensitivity of the measuring probe, it is necessary to choose a suitable method for implementation on the human body. In this section, the focus was on a flexible, easy, and comfortable way of implementation on the human body. For experimental measurements, two ways of implementing the measurement sensor were analyzed. These are implementation using a contact elastic strip (Figure 2(a)) or using a standard disposable hypoallergenic adhesive tape (Figure 2(b)), which is used for muscle and joint support and stability. The advantage of disposable hypoallergenic adhesive tape lies mainly in the very quick and easy implementation of the sensor on the human body. FIGURE 2. Placement of the measurement probe on the human body (shown in red) and how the sensor is implemented using an elastic strip (a) or hypoallergenic tape (b). B. FBG SENSOR ENCAPSULATION INTO ACRYLONITRILE BUTADIENE STYRENE (ABS) This section describes the complete process of encapsulating optical fiber with a Bragg grating into polymer layers during printing on a 3D printer. A large variety of materials can be used that differ in mechanical and thermal properties. ABS, which can withstand temperatures from −20 ◦C to 105 ◦C, was used to realize the prototype measurement probe. ABS exhibits a higher mechanical and lower temperature sensitivity compared to other commonly used materials such as PLA etc. Therefore, the ABS filament offers an optimal balance between strength and mechanical strain transfer while minimizing the effect of temperature fluctuations on the sensor output signal. These aspects are essential for this study because of the need to monitor very small vibrations caused by human heart activity and to reduce signal fluctuations due to changes in ambient temperature. The coefficient of thermal expansion of the ABS material ranges from 81 · 10−6to 95 ·10−6K−1, and the modulus of elasticity ranges from 2.0 GPa to 2.6 GPa. ABS is generally used for the manufacture of many products such as keyboard buttons, musical instrument components, medical devices, etc. The prototype sensor consists of a 10×5×2 mm main sensing section and an end section for fixing and protecting the lead fiber. A standard apodized Bragg grating in acrylic protection with a central Bragg wavelength of 1557 nm, a reflectance spectral width of 230 pm and a reflectivity of 90 % was used to realize the sensor. The process of manufacturing the prototype sensor was carried out in several steps. The Prusa i3 MK3S+ 3D printer (Prusa Research a.s., Prague, Czech republic) and ABS filament (Filament PM, Chudobin, Czech republic) with a diameter of 1.75 mm were used. First, the bottom part of the sensor was printed, which consisted of two layers with a thickness of 0.3 mm, and then the printing process was suspended. A bare optical fiber with a Bragg grating was placed on the bottom part so that the FBG in the primary protection was exactly in the middle of the sensor. The fiber was not glued to the bottom; it was only fixed with adhesive tape to the printing pad to prevent movement of the fiber when printing the sensor cover layer. This process was carried out very quickly to avoid significant cooling of the bottom layer material, which could have compromised the quality of the bond between the two layers. This was followed by printing the top part of the sensor using two 0.3 mm thick layers. The thickness of the 0.3 mm layer was chosen with respect to the diameter of the filament in primary protection (250 µm) to prevent the filament from being torn off or deflected out of its central position when the nozzle passes over it. During printing, a 15% fill was chosen to achieve the necessary flexibility of the sensor itself. Subsequently, a 900 µm diameter protection tube was strung onto the optical fiber, followed by a shrink tube to ensure the protection tube was fixed to the sensor itself. The resulting sensor is shown in Figure 3(b). C. THERMAL SENSITIVITY COEFFICIENT The sensor temperature sensitivity was defined by measurements in a temperature box in a temperature range from - 20 ◦C to 50 ◦C. The dependence of the Bragg wavelength VOLUME 12, 2024 57153 M. Fajkus et al.: FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology FIGURE 3. Encapsulation of the sensor: (a) placement of the optical fiber with a Bragg grating on the bottom layer; (b) the resulting FBG sensor. on temperature is shown in Figure 4. A prototype sensor realized by encapsulating a Bragg grating with a wavelength of 1557 nm in ABS plastic shows a temperature sensitivity of 55.35 pm/◦C, more than five times that of a bare Bragg grating according to normalized coefficient in equation 3and in equation. 4. This temperature sensitivity was determined by a linear approximation of the measured data with a confidence value of R =0.9518. FIGURE 4. Temperature dependence of the Bragg wavelength of the prototype sensor with a marked linear and second-order polynomial approximation. Due to the temperature-dependent coefficient of thermal expansion, the dependence of the Bragg wavelength of the measuring probe on temperature is a second-order polynomial function with a confidence value of R =0.9968, so this dependence can be expressed by the following relation: λB= −6.749 ·10−4T2+0.07671T+λB0,(5) where Tis the applied temperature, λB0is the Bragg wavelength of the grating at room temperature and λBis the measured Bragg wavelength of the measuring probe at temperature T. The determined average temperature sensitivity of the sensor is 55.35 pm/◦C. This significant increase results from the differing thermal expansions of the materials (glass fiber, acrylic protection, ABS filament) and the adhesion between these materials, leading to an increase in the non-linearity of the temperature characteristic of the proposed sensor. D. DEFORMATION SENSITIVITY COEFFICIENT The strain sensitivity was determined by gluing a prototype sensor using cyanoacrylate adhesive to a 30 ×27.8 ×1 mm steel plate. This plate was placed in a hydraulic press in which it was loaded with a tensile force of 0 to 1910 N. The deformation of the steel plate is given by: εsteel =F SE ,(6) where Fis the applied force, S is the cross section of the steel plate, and εsteel is the modulus of elasticity of the steel with a value of 210 GPa. The deformation of the prototype sensor can be determined by the following relationship: εFBG =1λB kε ,(7) where 1λBis the change in Bragg wavelength due to tensile loading and kεis the unknown deformation coefficient. This can be determined in two ways. First, equations 6and 7can be compared and the coefficient can be calculated according to the following relation: kε=1λBiSE Fi ,(8) where 1λBi represents the measured change in the Bragg wavelength for a given value of Fi. The calculated coefficients for each measured force are given in Table 2. The second method is based on an optimization problem in which the minimum error is sought between the calculated steel strain εsteel and the strain of the measuring probe εFBG in each of the n-measurements. The coefficient kεis then determined by solving the optimization problem by finding the minimum of the function f(kε)on the interval k∈⟨1;2⟩according to the following relation: f(kε)=Min n X i=1     Fi SE −1λBi kε     ,(9) where Firepresents the set load and kεis the searched deformation sensitivity parameter. From the above analysis, also shown in Figure 5, the deformation coefficient was defined at 1.249 pm/µstrain. The following table shows the values of the calculated mechanical stress and deformation of the steel plate, the measured deformation of the Bragg grating for different load values, and the calculated deformation coefficient at a given load step. The Bragg wavelength shifts during steel plate loading with the applied forces listed in Table 2are shown in 57154 VOLUME 12, 2024 M. Fajkus et al.: FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology TABLE 2. Calculated and measured values of stress, steel deformation, and prototype sensor. FIGURE 5. Determination of the deformation coefficient from the minimum error analysis. Figure 6(a). Figure 6(b) shows the dependence of the calculated steel strain and the measured Bragg grating strain on the applied force using a strain sensitivity factor of 1.249 pm/µstrain. The results show that the measuring probe with a Bragg grating encapsulated in ABS plastic exhibits a very good linear sensitivity to tensile stress. E. SIGNAL PROCESSING This chapter focuses on the analysis and interpretation of data obtained from Bragg grating-based sensors described in the previous chapters, describing the use of an interrogator unit to record the data, methods for filtering out noise, techniques for detecting peaks corresponding to heart rate, and procedures for smoothing the heart rate curve. The FBGuard interrogator unit (Safibra s.r.o., Czech Republic), which uses the principle of spectral analysis of reflected light from Bragg gratings, was used to record data from the FBG sensors. Data recording was performed at a sampling rate of 1000 samples/s. The measurement scheme is shown in Figure 7(a). Unwanted noise in the signal (muscle activity, motion artifacts) is created by higher frequencies, which are filtered out by a third-order digital Butterworth band-pass filter with cutoff frequencies of 1 and 15 Hz. Due to the lower cutoff frequency of 1 Hz, the fluctuation of the mean signal value caused by the varying depth of breathing of the test subjects over time is filtered out. The upper limit frequency was set at 15 Hz. At this cut-off frequency, sharpening of significant FIGURE 6. Dependence of the Bragg wavelength shift during loading of the steel plate (a); Dependence of the deformation of the steel and the Bragg grating on the applied force (b). parts was ensured, which is important for the subsequent detection of beats during cardiac activity. Subsequently, the signal is normalized and centered at zero mean. Peaks corresponding to heart rate (HR) are detected above this signal, from which the heart rate is calculated using the relation HR =60/(tn−tn−1), where tnis the time stamp of the n-th peak and tn−1is the time stamp of the preceding peak. The following is the process of smoothing the heart rate curve. A median filter with a window size of 7 is used for smoothing, which is statistically more robust to outlying observations than the moving average; see Figure 7(b). F. VERIFICATION OF SENSOR FUNCTIONALITY—HR MONITORING Measurements were performed under laboratory conditions on 11 healthy test subjects with their written informed VOLUME 12, 2024 57155 M. Fajkus et al.: FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology FIGURE 7. Basic schematic diagram of signal processing (a); Signal processing from the FBG sensor for HR determination (b). consent. The test subjects were 6 male (M1 to M6) and 5 female (F1 to F5). The age of the test subjects ranged from 21 to 47 years, height from 154 to 189 cm, and weight from 48 to 96 kg. The test was performed in the supine body position to simulate the MRI and sleep laboratory environment. Each test subject was asked to simulate different breathing loads, i.e. deep and shallow breathing, during the measurements. A MIKROE-2455 3-lead ECG (Mikroe, Serbia) was used for the reference measurements. 1) SIGNAL WAVEFORMS DURING SIGNAL PROCESSING In the following section, the waveforms of the measured signals and the influence of signal processing on these waveforms are shown for the selected subject M4. Figure 8shows ten-second examples of signals from FBG A (sensor fixed with an elastic strip) and FBG B (sensor fixed with adhesive tape). For better readability of the sensor response to respiratory activity, the mean value is subtracted from the signal. At first glance, a sensor fixed with adhesive tape shows approximately 5 times greater response to human respiratory activity than a sensor fixed with an elastic band. Both fiber optic sensors record respiratory activity (breathing) and cardiac activity due to chest expansion. The signal corresponding to cardiac activity has a significantly smaller response and is modulated to the slow component of the signal corresponding to respiratory activity. Figure 9shows the signal from the reference ECG device; in this case, no filtering is needed, only the R wave is detected. To remove the DC component and filter out the slow components in FBG sensors that correspond to respiratory activity, a third-order Butterworth bandpass filter with cutoff frequencies of 1 and 15 Hz is applied to the measured signals. The filtered signal from FBG A is shown in Figure 10(a), and the filtered signal from FBG B is shown in Figure 10(b). For both, the detected peaks corresponding to cardiac activity are FIGURE 8. Detail of the signal measured from the sensors (a) FBG A fixed with elastic strip and (b) FBG B fixed with adhesive tape. FIGURE 9. Signal from the reference ECG. marked. Peak detection in the signal was carried out in the MATLAB environment using the findpeaks function, with the MinPeakProminence and MinPeakDistance carefully adjusted for each subject tested to achieve the most accurate estimate of cardiac activity. Figure 11 shows a comparison of the 10-second signals from FBG sensor A fixed with an elastic band and the ECG signal. It can be seen from the figure that the detection of cardiac activity by the FBG sensor is fully correlated with the reference ECG signal. The waveforms also show a lower signal-to-noise ratio for the FBG sensor, which is due to the bandpass filtering of the signals (1 to 15 Hz) and thus the reduced bandwidth of the FBG sensor. However, the results described in the next section demonstrate that the FBG 57156 VOLUME 12, 2024 M. Fajkus et al.: FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology FIGURE 10. Detail of the signal measured from the sensors (a) FBG A fixed with elastic strip and (b) FBG B fixed with adhesive tape. sensor is suitable as an alternative method to monitor cardiac activity. The filtered signal from the FBG has a significantly smaller signal-to-noise ratio and also a smaller bandwidth due to filtering compared to the signal from the reference ECG. However, even in this way individual heartbeats are well detected, as shown in Figure 11. FIGURE 11. Comparison of waveforms of the FBG sensor A fixed with an elastic band and the reference ECG signal. From the time stamps of each peak, the heart rate is calculated using the formula HRi=60/(ti−ti−1), where HRicorresponds to the heart rate, tiis the timestamp of the i-th peak and ti−1) is the timestamp of the preceding peak. The calculated heart rate values for both sensors are compared with the calculated heart rate from the reference FIGURE 12. Comparison of HR from the fiber optic sensor and the reference ECG for (a) FBG A and (b) FBG B. FIGURE 13. Smoothed HR waveforms using median filter for (a) FBG A and (b) FBG B. ECG sensor and shown in Figure 12(a) for the FBG A sensor and Figure 12(b) for the FBG B sensor. VOLUME 12, 2024 57157 M. Fajkus et al.: FBG Sensor for Heart Rate Monitoring Using 3D Printing Technology TABLE 3. HR errors for individual subjects and sensors. FIGURE 14. Bland-Altman plot of the sensor success rate in HR measurement of subject M6 using (a) FBG A and (b) FBG B. To eliminate noise, the signal was further smoothed using a median average with a window size of 7 samples; see Figure 13. All measured data were processed in a similar manner and subjected to further analysis, which is described in the following section. 2) SENSOR COMPARISON IN THE TIME DOMAIN Based on the processing of the filtered signals and the HR calculation, the analysis of the results was performed and is summarized in Table 3. For each subject and sensor, the number of detected peaks N and the subsequently calculated heart rate variance σ2according to equation 10 are given: σ2=1 N N X i=1HRFBG,i−HRECG,i2,(10) where HRFBG,nis the n-th HR value calculated from the FBG fiber optic sensor and HRECG,nje n-th HR value calculated from the reference ECG device. The table also shows the average value of the heart rate error Ecalculated according to equation 11: E=1 N N X i=1 HRFBG,i−HRECG,i,(11) and the interval of these errors according to equation 12. EI =max (HRFBG −HRECG)−min (HRFBG −HRECG). (12) 3) COMPARISON OF SENSORS USING BLAND-ALTMAN ANALYSIS For two selected subjects, Figures 14 and 15 show B-A plots that graphically represent the sensor success rate compared to a reference ECG sensor. In Figure 14, the B-A analysis of the selected male subject M6 is shown, where 92.28% of all samples lie within ±1.96 SD (Standard Deviation) for the FBG A sensor, see Figure 14(a) and 95.9% for the FBG B sensor, see Figure 14(b). Similarly, the Bland-Altman analysis for the female subject F2 is depicted graphically; in the case of FBG A, 95.11% of all samples lie within ±1.96 SD, see Figure 15(a), while in the case of FBG B, a total of 95.15% of samples lie within ±1.96 SD, see Figure 15(b). A statistical summary of the collected data is shown in Table 4. The first column represents the subject designation, and the second column indicates the type of Bragg sensor (FBG A sensor fixed with elastic strip, FBG B sensors fixed with adhesive tape). The other columns are related to the Bland-Altman (B-A) statistic. MoD represents the 57158 VOLUME 12, 2024