Performance of blind source separation (BSS) techniques for mixed source signals of EEG, ECG, and voice signal
Alvin Sahroni, Hendra Setiawan, Erlina Marfianti · 2014
This paper presents a study of Blind Source Separation (BSS) application in telemedicine problem, especially during medical data recording. There are two techniques that will be investigated, Natural Gradient Method (NGM) and Fast Independent Coeffient Analysis(FastICA) with main source signals are Electrocardiograph (ECG), Electroencephalograph (EEG), and human voice signals. This study related the needs of doctor and patient to communicate each other in separate places, and also the doctor will be able to monitoring the EEG and ECG signals simultaneously during a call with the patients using a portable/mobile device that have been attached with additional module. Hopefully, while implementing BSS technique into the additional module for signal processing purpose, will increase the quality of medical outpatient system remotely. This paper reported that the performance of FastICA method is better about 80% effectively than NGM while separating mixed source signals of EEG, ECG, and voice Signal.