Monitoring Of Heart Activity Research Articles

Heart disease detection using an acceleration-deceleration curve-based neural network with consumer-grade smartwatch data

Cardiovascular disease (CVD) is the most important cause of morbidity and mortality worldwide. Early detection, prevention or even prediction is of pivotal importance to reduce the burden of cardiovascular disease and its associated costs. Low cost, consumer-grade smartwatches have the potential to revolutionize cardiovascular medicine by enabling continuous monitoring of heart rate and activity. When combined with machine learning(ML), the resulting large amounts of time series data hold the potential of detection, or exclusion of CVD. However, analyzing such large datasets is challenging due to the sparse presence of informative segments. Efficient selection of these segments is essential for developing predictive models for clinical deployment. The objective of this paper was to investigate the potential of an acceleration-deceleration curvebased ML model as a novel clinical indicator for the detection of cardiovascular diseases. We used data from the ME-TIME study; 42 participants from which 21 have a cardiovascular disease and 21 are health controls. Data from each subject was normalized to decrease inter-subject variability. A neural network model aggregated predictions per week. We showed that per-subject normalization by the peak value of curves during inactivity, aggregation of model predictions over a week, and using a contrastive loss, resulted in a predictive model with 99 % ± 3 % specificity and 40 % ± 49 % sensitivity on the development set, and 100 % specificity with 67 % ± 47 % sensitivity on the test set. Acceleration-deceleration curves are effective patterns for ruling out the presence of cardiovascular disease, but caution must be taken to properly pre-process the curves and carefully choosing a model that reduces the variability in the extracted curves.

Heartbeat classification using various machine learning models: A comparative study

Cardiac arrhythmias, known as irregular heartbeats, pose a notable health threat that necessitates prompt diagnosis, as untreated arrhythmias can lead to severe heart complications. Among the various methods for arrhythmia detection, electrocardiography is the most prevalent due to its non-invasive monitoring of heart activity. However, manual electrocardiogram (ECG) analysis is inefficient and prone to errors, prompting the exploration of machine learning (ML) models for ECG feature recognition. Integrating ML models with ECG analysis can revolutionize cardiac diagnostics by improving healthcare efficiency and outcomes by enhancing the accuracy and consistency of existing approaches as well as their processing speed for large datasets. Unfortunately, current ML methods encounter two key limitations: prolonged training times and the need for manual feature selection. To address these issues, we propose using ML models enhanced with innovative techniques such as the Fourier transform (FT) and Gaussian noise injection for improved cardiac health assessment. To validate this approach, we utilized statistical tools, including Pearson correlation and p-values, to uncover relationships within the data. In addition, we employed the FT technique to extract and analyze frequency-domain features. Our comparative study of different ML models relied on metrics such as accuracy, precision, recall, F1 score, and receiver operating characteristic area under the receiver operating characteristic curve, demonstrating XGBoost’s impressive average recall of 0.956 with 99.96% overall accuracy. An average precision of 0.956 further underscored the accuracy of XGBoost’s predictions, indicating its high level of reliability in distinguishing various cardiac conditions. These results highlight the considerable potential of ML techniques for precise ECG-based clinical diagnoses, helping healthcare professionals make more accurate and timely decisions in patient care.

Modification of SMWT for remote heart’s functional capacity determination from oxygen consumption rate

Cardiovascular disease (CVD) is series of diseases affects blood circulation and heart which are fatal to human being. CVD patients recorded that every each of them must have at least one among the following behavioural risks: smoking, diabetes, obesity, hypertension, high cholesterol, and sedentary lifestyle. The diagnosis for CVD is commonly through cardiac stress test. It is a test where electrocardiogram will be used to monitor heart activity under induced stress such as walking on treadmill. However, the facilities for the cardiac stress test are very limited and increasing number of patients with potential CVD caused a long queue for an appointment. This situation makes it hard for healthy people living with behavioral risk wanting for their heart to be checked. Therefore, in this paper, we proposed to determine heart functional capacity which can be done anywhere and sufficient to access heart condition. For this proposal, despite using treadmill-test, Six-Minutes Walking Test (SMWT) is utilized. SMWT induces stress while the heart’s endurance parameters such as heart rate is extracted. This study developed an integrated sensor device which comprises of a sensor to detect step count calculate distance, and a pulse sensor that detect heart rate. Then, oxygen consumption rate during the SMWT, is computed. Metabolic equivalent of task (MET) is equal to the oxygen consumption rate during activity over resting oxygen consumption rate. Heart functional capacity classification is about exercise intensity and the MET of SMWT can be compared to the classification which allows this paper to determine heart functional capacity remotely using SMWT. Based on the results, the heart’s functional capacity classification after performing SMWT using sensor device is Class 1, the same when using manual calculation for distance and heart rate. The device sensor somehow produce error at 20.63% when calculating oxygen uptake. Nevertheless, this paper is identified that the SMWT is able to classify the heart’s functional capacity.

Design a Simple Portable Electrocardiograph Based on the IOT Concept

Abstract There are many cases of heart disease with death associated. The symptoms are not always clearly visible and require examination using an electrocardiogram (ECG) device to be detected. Recording human electrical activity can be done in various organs of the human body, one of which is in the heart. An ECG is a medical test to detect electrical activity signals produced by the heart by exchanging the signals on a monitor or graph on paper. This study aimed to design, build, test, and analyze a portable ECG device that is compact and easy to use. This portable electrocardiogram helps monitor human heart rate to check heart rate periodically. Tool design is done by making tools equipped with a monitoring system. The proposed system uses the AD8232 sensor to detect a person’s cardiac activity. Arduino Nano is a microcontroller connected to ESP32 with a WiFi feature. Heart activity monitors can be done anywhere by applying the Internet of Things concept, namely with the Ubidots application. A portable tool using Li-Ion batteries as a tool voltage source and BMS 3S as a charger, this tool is also equipped with a 3.2-inch TFT screen as a graph viewer of the results of monitoring heart activity. The results showed that the system could monitor the user’s heart for 5 minutes when sitting, walking, and running. The system created has been able to send heart signal data to be processed by Arduino Nano and ESP32 as a data transfer tool so that the data generated in the form of electrocardiogram graphs can be displayed on the Ubidots application. TFT LCDs can display electrocardiogram graphs generated by the electrical activity of the human heart and beats per minute values. The AD8232 sensor can measure beat per minute (bpm) values with an average increase in bpm values for walking conditions of 17% and running conditions of 19.17%. The AD8232 Sensor Performance can detect the user’s heart rate with a percentage of data difference between walking and running conditions of 2.88% with an accuracy rate of 97.12%.

Cloud-Connected Patch-Worn Auscultation Device for Chest Sound Monitoring.

AbstractIn recent years, wearable electrocardiograms have risen in popularity as a solution for personal monitoring of heart activity. However, this technology has limitations in diagnostic capability and structural function monitoring. Meanwhile, auscultation of the heart remains a fundamental tool for physicians in diagnosis and monitoring of heart disease largely unaddressed in a convenient wearable format. The present work outlines a promising system currently under investigation, allowing user-initiated 10-second chest-sound recordings to be transmitted over Bluetooth-Low-Energy, with an innovative package design providing inherent noise reduction and a high signal-to-noise ratio. The device has been tested on healthy individuals, and system response has been validated against calibrated electrocardiogram recording equipment to analyze signal capture fidelity.Abstract Figure

Portable ECG monitoring system with IOT analysis

This system is designed to calculate the heart rate of patients and transmit the data in beats per minute (bpm) to a cloud database. By using this system, doctors can analyze critical parameters remotely. The project involves interfacing the AD8232 ECG Sensor with the NodeMCU ESP8266 Board to monitor ECG waveforms on a Serial Plotter Screen and transmit the data to an IoT cloud platform for real-time monitoring. This innovation allows for remote monitoring of heart activity, reducing the need for hospital stays. Utilizing the Blynk IoT platform, the system ensures online access via PC or Smartphone, representing a significant advancement in patient health monitoring. This approach aims to improve early detection and prevention of heart diseases, enhancing patient care and reducing fatalities.

Novel Monitoring and Treatment Technologies for the Heart.

Cardiovascular disease may be the world's leading killer of men and women, but new technologies are in development that could help lessen its impact. Among them are a variety of innovative external and internal patches that employ flexible and stretchable materials, machine learning, and other tactics to monitor heart activity and function, and in some cases to provide on-the-spot treatment.

Enabling Complex Impedance Spectroscopy for Cardio-Respiratory Monitoring with Wearable Biosensors: A Case Study

Recent advances in commercially available integrated complex impedance spectroscopy controllers have brought rapid increases in the quality of systems available to researchers for wearable and remote patient monitoring applications. As a result, novel sensing methods and electrode configurations are increasingly viable, particularly for low-power embedded sensors and controllers for general electrochemical analysis. This study evaluates a case study of the four electrode locations suitable for wearable monitoring of respiratory and heart activity monitoring using complex impedance spectroscopy. We use tetrapolar electrode configurations with ten stimulation frequencies to characterize the relative differences in measurement sensitivity. Measurements are performed and compared for the magnitude, phase, resistive, and reactive components of the bioimpedance using two COTS-based controllers, the TI AFE4300 and MAX30009. We identify the highest percent relative changes in the magnitude of the impedance corresponding to deep breathing and heart activity across the chest (17% at 64 kHz, 0.5% at 256 kHz, respectively), on the forearm (0.098% at 16 kHz, 0.04% at 8 kHz), wrist-to-wrist across the body (0.28% at 256 kHz, 0.04% at 256 kHz, respectively), and wrist-to-finger across the body (0.35% at 4 kHz, 0.05% at 4 kHz, respectively). We demonstrate that the wrist-to-wrist and wrist-to-finger configurations are most promising and may enable new wearable bioimpedance applications. Additionally, this paper demonstrates that deep respiration and heart activity influence bioimpedance measurements in whole-body measurement configurations, with variations of nearly 1% in measured impedance due to the phase of the breathing cycle.

Textile-based Wearable to Monitor Heart Activity in Paediatric Population: A Pilot Study.

Cardiac monitoring for children with heart disease still employs common clinical techniques that require visits to hospital either in an ambulatory or inpatient setting. Frequent cardiac monitoring, such as heart rate monitoring, can limit children's physical activity and quality of life. The main objective of this study is to evaluate the performance of a textile-based device (SKIIN) in measuring heart rate (HR) in different tasks: lying down, sitting, standing, exercising, and cooling down. Twenty participants including healthy children and children with heart disease were included in this study. The difference between the HRs recorded by the SKIIN was compared with a reference electrocardiogram collection by normalized root mean squared error. Participants completed a questionnaire on their experience wearing the textile device with additional parental feedback on the textile device collected. Participants had the median age of 14 years (range: 10-17 years), with body mass index 23.1 ± 3.8 kg/m2 and body surface area 1.70 ± 0.25 m2. The HR recorded by SKIIN and reference system significantly changes between tasks (P < 0.001), while not significantly different from each other (P > 0.05). The normalized root mean squared error was 3.8% ± 3.0% and 3.6% ± 3.7% for healthy and the heart disease groups, respectively. All participants found the textile device non-irritating and easy to wear. This study provides proof of concept that HR can be robustly and conveniently monitored by smart textiles, with similar accuracy to standard-of-care devices.

Continuous monitoring of patients in and after the acute admission ward to improve clinical pathways: study protocol for a randomized controlled trial (Optimal-AAW)

BackgroundBecause of high demand on hospital beds, hospitals seek to reduce patients’ length of stay (LOS) while preserving the quality of care. In addition to usual intermittent vital sign monitoring, continuous monitoring might help to assess the patient’s risk of deterioration, in order to improve the discharge process and reduce LOS. The primary aim of this monocenter randomized controlled trial is to assess the effect of continuous monitoring in an acute admission ward (AAW) on the percentage of patients who are discharged safely.MethodsA total of 800 patients admitted to the AAW, for whom it is equivocal whether they can be discharged directly after their AAW stay, will be randomized to either receive usual care without (control group) or with additional continuous monitoring of heart rate, respiratory rate, posture, and activity, using a wearable sensor (sensor group). Continuous monitoring data are provided to healthcare professionals and used in the discharge decision. The wearable sensor keeps collecting data for 14 days. After 14 days, all patients fill in a questionnaire to assess healthcare use after discharge and, if applicable, their experience with the wearable sensor. The primary outcome is the difference in the percentage of patients who are safely discharged home directly from the AAW between the control and sensor group. Secondary outcomes include hospital LOS, AAW LOS, intensive care unit (ICU) admissions, Rapid Response Team calls, and unplanned readmissions within 30 days. Furthermore, facilitators and barriers for implementing continuous monitoring in the AAW and at home will be investigated.DiscussionClinical effects of continuous monitoring have already been investigated in specific patient populations for multiple purposes, e.g., in reducing the number of ICU admissions. However, to our knowledge, this is the first Randomized Controlled Trial to investigate effects of continuous monitoring in a broad patient population in the AAW.Trial registrationhttps://clinicaltrials.gov/ct2/show/NCT05181111. Registered on 6 January 2022. Start of recruitment: 7 December 2021.

An Arrhythmia Classification Model Based on Vision Transformer with Deformable Attention.

The electrocardiogram (ECG) is a highly effective non-invasive tool for monitoring heart activity and diagnosing cardiovascular diseases (CVDs). Automatic detection of arrhythmia based on ECG plays a critical role in the early prevention and diagnosis of CVDs. In recent years, numerous studies have focused on using deep learning methods to address arrhythmia classification problems. However, the transformer-based neural network in current research still has a limited performance in detecting arrhythmias for the multi-lead ECG. In this study, we propose an end-to-end multi-label arrhythmia classification model for the 12-lead ECG with varied-length recordings. Our model, called CNN-DVIT, is based on a combination of convolutional neural networks (CNNs) with depthwise separable convolution, and a vision transformer structure with deformable attention. Specifically, we introduce the spatial pyramid pooling layer to accept varied-length ECG signals. Experimental results show that our model achieved an F1 score of 82.9% in CPSC-2018. Notably, our CNN-DVIT outperforms the latest transformer-based ECG classification algorithms. Furthermore, ablation experiments reveal that the deformable multi-head attention and depthwise separable convolution are both efficient in extracting features from multi-lead ECG signals for diagnosis. The CNN-DVIT achieved good performance for the automatic arrhythmia detection of ECG signals. This indicates that our research can assist doctors in clinical ECG analysis, providing important support for the diagnosis of arrhythmia and contributing to the development of computer-aided diagnosis technology.

Multilayer Flexible SU8-Gold Microelectrode Arrays for Wearable Bioelectronics

Wearable health trackers for vital signs monitoring are becoming ever more important especially due to the global coronavirus pandemic (COVID-19) caused by the SARS‑CoV‑2 virus which severely affect the respiratory system and can cause cardiac manifestations. Particularly, wearable solutions which can seamlessly monitor heart activity are critical to facilitate personal preventive and remote healthcare, as well as to allow early diagnosis of cardiac dysfunctions. A fundamental enabler of wearable bioelectronics is the sensing bioelectrode which is used to record surface biopotentials. While a plethora of attempts have been reported to realize skin-conformal dry electrodes and electronic skin patches, oftentimes a very critical aspect of the electrode i.e., the actual electrical interfacing of the wearable electrode to readout circuits without disturbing the skin-electrode contact, is overlooked. To address this issue, this paper reports a unique tri-layer, polymer-metal-polymer skin-conformal microelectrode design with sidewall metal coating to achieve vertical interconnect accesses (VIAs) and realize contact pads for external interfacing. The novel and optimized process flow reported herein allows repeatable fabrication of flexible electrodes in arrayed format with yields exceeding 90%. Functionality of the microfabricated electrodes were demonstrated by successful acquisition of the electrocardiogram in lead-I configuration with clear detection of the P-QRS-T complex.

Severity-Based Hierarchical ECG Classification Using Neural Networks.

Timely detection of cardiac arrhythmia characterized by abnormal heartbeats can help in the early diagnosis and treatment of cardiovascular diseases. Wearable healthcare devices typically use neural networks to provide the most convenient way of continuously monitoring heart activity for arrhythmia detection. However, it is challenging to achieve high accuracy and energy efficiency in these smart wearable healthcare devices. In this work, we provide architecture-level solutions to deploy neural networks for cardiac arrhythmia classification. We have created a hierarchical structure after analyzing various neural network topologies where only required network components are activated to improve energy efficiency while maintaining high accuracy. In our proposed architecture, we introduce a severity-based classification approach to directly help the users of the wearable healthcare device as well as the medical professionals. Additionally, we have employed computation-in-memory based hardware to improve energy efficiency and area consumption by leveraging in-situ data processing and scalability of emerging memory technologies such as resistive random access memory (RRAM). Simulation experiments conducted using the MIT-BIH arrhythmia dataset show that the proposed architecture provides high accuracy while consuming average energy of 0.11 μJ per heartbeat classification and 0.11 mm2 area, thereby achieving 25× improvement in average energy consumption and 12× improvement in area compared to the state-of-the-art.

Automatic varied-length ECG classification using a lightweight DenseNet model

Atrial fibrillation is the most common abnormal heart condition and contributes primarily to cardiac morbidity and mortality. In the last decades, portable Electrocardiogram (ECG) monitor has effectively supported patients with periodic heart arrhythmia by constantly monitoring heart activity and early detecting heart arrhythmia. However, diagnosing the ECG recordings, which is a demanding and time-consuming job, is mainly relied on experienced medical workers. It is of great importance to develop methods to assist in diagnosing ECG recordings. In this work, we propose a lightweight model architecture based on the Densely Connected Convolutional Networks to classify cardiac arrhythmia automatically. The Cyclical Learning Rate was utilized to automatically change the learning rate value throughout the training procedure. Our model was trained and tested on the PhysioNet/Computing in Cardiology Challenge 2017 (CinC2017) and 1st China Physiological Signal Challenge 2018 (ICBEB2018) datasets. The obtained results have been compared with other works using metric F1 for CinC2017 and metrics F1,AUC,Fmax,Fβ=2,Gβ=2 for ICBEB2018. Though our structure consists of a low number of parameters and requires less computational costs, its performance is in line with the state-of-the-art Deep Neural Networks with F1 score of 0.831 and 0.826 for CinC2017 and ICBEB2018 datasets, respectively.

Simultaneous screening of zebrafish larvae cardiac and respiratory functions: a microfluidic multi-phenotypic approach.

Multi-phenotypic screening of multiple zebrafish larvae plays an important role in enhancing the quality and speed of biological assays. Many microfluidic platforms have been presented for zebrafish phenotypic assays, but multi-organ screening of multiple larvae, from different needed orientations, in a single device that can enable rapid and large-sample testing is yet to be achieved. Here, we propose a multi-phenotypic quadruple-fish microfluidic chip for simultaneous monitoring of heart activity and fin movement of 5-7-day postfertilization zebrafish larvae trapped in the chip. In each experiment, fin movements of four larvae were quantified in the dorsal view in terms of fin beat frequency (FBF). Positioning of four optical prisms next to the traps provided the lateral views of the four larvae and enabled heart rate (HR) monitoring. The device's functionality in chemical testing was validated by assessing the impacts of ethanol on heart and fin activities. Larvae treated with 3% ethanol displayed a significant drop of 13.2 and 35.8% in HR and FBF, respectively. Subsequent tests with cadmium chloride highlighted the novel application of our device for screening the effect of heavy metals on cardiac and respiratory function at the same time. Exposure to 5$\mu$g/l cadmium chloride revealed a significant increase of 8.2% and 39.2% in HR and FBF, respectively. The device can be employed to monitor multi-phenotypic behavioral responses of zebrafish larvae induced by chemical stimuli in various chemical screening assays, in applications such as ecotoxicology and drug discovery.

PEDOT:PSS-modified cotton conductive thread for mass manufacturing of textile-based electrical wearable sensors by computerized embroidery

The textile industry has advanced processes that allow computerized manufacturing of garments at large volumes with precise visual patterns. The industry, however, is not able to mass fabricate clothes with seamlessly integrated wearable sensors, using its precise methods of fabrication (such as computerized embroidery). This is due to the lack of conductive threads compatible with standard manufacturing methods used in industry. In this work, we report a low-cost poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS)-modified cotton conductive thread (PECOTEX) that is compatible with computerized embroidery. The PECOTEX was produced using a crosslinking reaction between PEDOT:PSS and cotton thread using divinyl sulfone as the crosslinker. We extensively characterized and optimized our formulations to create a mechanically robust conductive thread that can be produced in large quantities in a roll-to-roll fashion. Using PECOTEX and a domestic computerized embroidery machine, we produced a series of wearable electrical sensors including a facemask for monitoring breathing, a t-shirt for monitoring heart activity and textile-based gas sensors for monitoring ammonia as technology demonstrators. PECOTEX has the potential to enable mass manufacturing of new classes of low-cost wearable sensors integrated into everyday clothes.

An examination into the mental and physical effects of a saffron extract (affron®) in recreationally-active adults: A randomized, double-blind, placebo-controlled study.

ABSTRACTBackgroundSaffron, derived from the stigmas of the Crocus Sativus flower, has been shown in several studies to improve mood and wellbeing in adults experiencing low mood and anxiety. The goals of this study were to examine its mental and physical effects in healthy, recreationally active adults.MethodsIn this 6-week, randomized, double-blind, placebo-controlled study, 62 adults engaging in regular exercise were recruited and randomized to receive a placebo or 28 mg daily of a standardized saffron extract (affron®). Self-report outcome measures include the Physical Activity Enjoyment Scale, Profile of Mood States, and Patient-Reported Outcomes Measurement Information System-29. Participants also wore a wrist-worn heart rate, activity, and sleep monitoring device (WHOOP) to measure changes in sleep quality, resting heart rate, and heart rate variability. To help identify mechanisms of action associated with saffron intake, changes in plasma concentrations of brain-derived neurotrophic factor, oxytocin, and neuropeptide Y were also measured.ResultsBased on data collected from all participants, there were no statistically significant between-group differences in changes in any of the outcome measures. However, when changes were analyzed by sex, there were statistically significant greater increases in enjoyment associated with exercise (p =.009) and heart rate variability (p =.001) in male participants taking saffron compared to the placebo. No statistically significant between-group differences were identified in females.ConclusionsThe results of this trial suggest saffron may have beneficial effects in recreationally active males, as evidenced by increased exercise enjoyment and heart rate variability. However, no such benefits were identified in females. Future research using larger sample sizes, varying treatment periods, and additional outcome measures will be required to validate the results from this study and help clarify the mechanisms of action associated with saffron intake.This study was prospectively registered on 30 October 2020 with the Australian and New Zealand Clinical Trials Registry (Trial ID. ACTRN12621000501842).

Ultrasoft Porous 3D Conductive Dry Electrodes for Electrophysiological Sensing and Myoelectric Control.

Biopotential electrodes have found broad applications in health monitoring, human-machine interactions, and rehabilitation. Here, we report the fabrication and applications of ultrasoft breathable dry electrodes that can address several challenges for their long-term wearable applications - skin compatibility, wearability, and long-term stability. The proposed electrodes rely on porous and conductive silver nanowire based nanocomposites as the robust mechanical and electrical interface. The highly conductive and conformable structure eliminates the necessity of conductive gel while establishing a sufficiently low electrode-skin impedance for high-fidelity electrophysiological sensing. The introduction of gas-permeable structures via a simple and scalable method based on sacrificial templates improves breathability and skin compatibility for applications requiring long-term skin contact. Such conformable and breathable dry electrodes allow for efficient and unobtrusive monitoring of heart, muscle, and brain activities. In addition, based on the muscle activities captured by the electrodes and a musculoskeletal model, electromyogram-based neural-machine interfaces were realized, illustrating the great potential for prosthesis control, neurorehabilitation, and virtual reality.

Breathing Rate Estimation from Head-Worn Photoplethysmography Sensor Data Using Machine Learning.

Breathing rate is considered one of the fundamental vital signs and a highly informative indicator of physiological state. Given that the monitoring of heart activity is less complex than the monitoring of breathing, a variety of algorithms have been developed to estimate breathing activity from heart activity. However, estimating breathing rate from heart activity outside of laboratory conditions is still a challenge. The challenge is even greater when new wearable devices with novel sensor placements are being used. In this paper, we present a novel algorithm for breathing rate estimation from photoplethysmography (PPG) data acquired from a head-worn virtual reality mask equipped with a PPG sensor placed on the forehead of a subject. The algorithm is based on advanced signal processing and machine learning techniques and includes a novel quality assessment and motion artifacts removal procedure. The proposed algorithm is evaluated and compared to existing approaches from the related work using two separate datasets that contains data from a total of 37 subjects overall. Numerous experiments show that the proposed algorithm outperforms the compared algorithms, achieving a mean absolute error of 1.38 breaths per minute and a Pearson’s correlation coefficient of 0.86. These results indicate that reliable estimation of breathing rate is possible based on PPG data acquired from a head-worn device.

Fast Skin Segmentation on Low-Resolution Grayscale Images for Remote PhotoPlethysmoGraphy

Facial skin segmentation is an important preliminary task in many applications, including remote PhotoPlethysmoGraphy (rPPG), which is the problem of estimating the heart activity of a subject just by analyzing a video of their face. By selecting all the subject’s skin surface, a more robust pulse signal could be extracted and analyzed in order to provide an accurate heart activity monitoring. Single-photon avalanche diode cameras have proven to be able to achieve better results in rPPG than traditional cameras. Although this kind of cameras produces accurate photon counts at high frame rate, they are able to capture just grayscale low resolution images. For this reason, in this work, we propose a novel skin segmentation method based on deep learning that is able to precisely localize skin pixels inside a low-resolution grayscale image. Moreover, since the proposed method makes use of depthwise separable convolutional layers, it could achieve real-time performances even when implemented on a small low-powered IoT device.