Unconstrained ECG monitoring systems

Abstract

Electrocardiogram (ECG) records the electrical activity of the heart in a noninvasive manner with high temporal resolution. Long-term daily monitoring via ECG is important for assessing and preventing sleep-related disorders as well as cardiac-related diseases. There are three major types of sensing methods for ECG recordings. Choosing a suitable sensing technology according to ambient conditions enables more practical, user-friendly, and unconstrained ECG monitoring. In this dissertation, two ECG sensing technologies were selected to accommodate different amounts of movement, and were evaluated for unconstrained ECG monitoring both during sleep and in daily activities. In the first study, we developed and tested an unconstrained and non-contact ECG measurement system using a capacitively-coupled (CC) electrode array on a bed for long-term monitoring of heart activity during sleep. The system was composed of multi-channels to increase the ECG signal quality in various sleep positions and postures in a bed. Thirteen healthy subjects participated in the experiment, and the system had a high sensitivity of R-peak detection (97%) and very low root mean square errors (RMSEs) of RR intervals and heart rate (HR) (1.46 ms and 0.09 bpm). The performance of the automatic channel selection algorithm based on an autocorrelation function, power spectrum, and kurtosis showed an accuracy of 99.54%. Moreover, the measured ECG signals were used to classify four sleep postures, supine, right lateral, prone, and left postures, in a bed, based on the change in ECG waveform according to the body posture. The features were applied to linear discriminant analysis, support vector machines (SVMs) with linear and radial basis function (RBF) kernels, and artificial neural networks (ANNs), respectively. SVM with RBF kernel demonstrated the highest performance with an accuracy of 98.4% for estimation of four sleep postures. Overall, although ECG data were obtained from only a few sensors in an unconstrained manner, the method outperformed other results that have been reported to date. The developed system and algorithm can be applied to ECG monitoring during sleep as well as the management of sleep parameters related to obstructive sleep apnea and bedsores. In the second study, the conventional rigid and hard type CC electrodes were replaced with conductive textile electrodes for long-term cardiac monitoring and user-friendly applications during sleep. Therefore, we developed and tested an unconstrained and non-contact mattress type 1-channel ECG measurement system based on CC textile electrodes on a bed. The system, which was designed to measure ECG in a bed with no constraints regarding sleep position and posture, included a foam layer to increase the contact region with the curvature of the body and a cover to ensure durability and easy installation. Nine healthy subjects participated in the experiment during polysomnography. The experimental results showed that the sensitivity of R-peak detection was 98.0%, and the normalized errors of heart rate variability time and spectral measures ranged from 0.15%-4.20%. The RMSEs for RR intervals and HR were 1.36 ms and 0.09 bpm, respectively. The performances of this system exceeded those that have been reported to date. Furthermore, the measured ECG signals were applied to wake/sleep estimation based on body movement and HR variation. The accuracy and kappa of the epoch-by-epoch analysis were 88.7% and 0.59, respectively that were similar to the results obtained using actigraphy. Therefore, the developed system and algorithm have the potential for daily ECG monitoring and sleep quality monitoring in an unobtrusive way at home. In the third study, as techniques based on CC electrodes were highly vulnerable to motion artifacts, a method was proposed for developing an optimal electrode system for unconstrained ECG monitoring in daytime activities, in the form of a small and wearable single-patch ECG monitoring device that allows for the faithful reconstruction of the standard 12-lead ECG. The optimized universal electrode positions on the chest and the personalized transformation matrix were determined using linear regression as well as ANNs. A total of 24 combinations of 4 neighboring electrodes on 35 channels were evaluated on 19 healthy subjects. Moreover, we analyzed combinations of three electrodes within the four-electrode combination with the best performance. The mean correlation coefficients were all higher than 0.94 in the case of the ANN method for the combinations of four neighboring electrodes. The reconstructions obtained using the three and four sensing electrodes showed no significant differences. The reconstructed 12-lead ECG obtained using the ANN method is better than that obtained using the multilinear regression method. Therefore, three sensing electrodes and one ground electrode (forming a square) placed below the clavicle on the left were determined to suitable for ensuring good reconstruction performance. The results also showed that small unavoidable changes in the electrode positions did not affect reconstruction performance adversely. Finally, because the interelectrode distance was determined to be 5 cm, the suggested approach can be implemented in a single-patch device, which should allow for the continuous monitoring of the standard 12-lead ECG without requiring limb contact, both in daily life and in clinical practice. The experimental results demonstrated that the developed systems, the non-contact bed-type systems and wearable patch-type system, were improvements upon the existing technologies, and the performances of the proposed sleep posture classification, wake/sleep estimation, and 12-lead ECG reconstruction methods were comparable to those of previous studies. The systems and methods show promise for contributing to the daily monitoring of the heart and applications of ECG into daily activities as well as during sleep.