Research of Independent Component Analysis for Processing Bowel Sounds / 中国医学物理学杂志

ABSTRACT

Objective: To eliminate the noise merged in bowel sounds and extract the time-domain features of the main signal. Methods: In this paper, the principles of Independent Conponent Analysis (ICA) were investigated; the algorithm of Fast ICA was discussed; the step of ICA was given. Then we processed the bowel sounds with this method, and extracted the time-domain features with normalized average Shannon energy, representative differences were found between the normal and abnormal signals. Results: The noise merged in bowel sounds is eliminated and there are statistical differences between the normal and abnormal bowel sounds. Conclusions: Detection and analysis of bowel sounds has important value in the diagnosis and cure of gastrointestinal diseases, the experimental results show that ICA is an effective method for processing bowel sounds.