Remote signal detection and processing in a wearable ECG systems

The comprehension and correct monitoring of the electrocardiogram (ECG) signals are very important for the human health and continuous monitoring of chronically ill patients. For this purpose, there are several ongoing and completed studies on wearable monitoring systems. In this thesis, a wearable ECG system with real time signal processing and monitoring features was developed. The system, a conductive textile fabric produced by using nanotechnologies for very different purposes was used as electrodes. The usability and performances of conductive textile electrodes evaluated with disposable Ag-AgCl electrodes in R-R intervals of ECG signals. Textile electrodes gave approximately % 25,23 better results compared to Ag-AgCl electrodes. Textile electrodes were fixed to a wearable cloth that was designed in the form of corset together with ECG node. The designed ECG node is low power consumption microprocessor based and communicates over Bluetooth. ECG node spends approximately 35 mA at the time of communication. It has been observed in experiments performed with the system that, although the conductive fabric whose resistance is close to almost zero and sensing ECG very well, that it also identified environmental or physiological noises were detected. Therefore, a real time system that is compatible with in Android based systems was designed to clean different artifacts that makes the detection and follow-up of ECG signals difficult. The extended Kalman filter was used to remove the random noises such as EMG that was not able to be modeled exactly and motion artifacts. This suggested real-time software was compared with other methods in terms of signal-to-noise ratio (SNR) by using EMG recordings in MIT-BIH database. Then, it was evaluated with MSE, PRD, SNR parameters obtained from different sizes of EMG noises. On the recordings numbered 111, 113, 115, 121, 122 from MIT-BIH database, EMG noise was added in the form of SNR = 5 dB at the beginning. The SNR results obtained from ECG signs are respectively, 9.0348, 8.367, 7.67, 8.73, 8.72. EMG noise was added to the same data set in the form of SNR = 10 dB. The SNRs obtained from the algorithm under these conditions are respectively 10.65, 10.64, 8.95, 10.19, 18.10. In this thesis, UDP, TCP and Web Socket communication protocols have been tested. UDP has been the fastest method for the ECG signal transferring from the patient to the doctor. At the same time, a method is proposed for direct access to the patient by the doctor. Since the system will be used in the long-term follow-up as real-time, by considering the daily life situations of people, their working by "standing", "being on the computer", "lying" and "going up the stairs" were monitored. The obtained results show that proposed textile electrode and wearable system are suitable for long-term ECG monitoring. This type of systems will submit highly ergonomic solutions among biomedical device technologies. In addition, for athletes, for the patients who require long-term follow-up and for the disorders such as sleep apnea and the diagnosis of arrhythmias that is ambiguous about when it will come out, the usage of such kind of systems is foreseen.