Abstract
The small intestine plays a very important role in digestive absorption of food. Whether certain diets are good for physical conditions can be judged by bowel sounds. However, there are problems with using bowel sounds to judge food quality. In this paper we discuss methods for improving the judgment of food quality on the basis of bowel sounds. A high sensitivity acceleration sensor was placed on the surface of the abdomen in four places to detect bowel sounds. A multi-signal processing system consisting of A/D converter, DAT, HD and a personal computer was used for the data analysis. Independent component analysis (ICA) for pretreatment of data signals and application of six ICA algorithms improved the accuracy of bowel sound detection. The results of this analysis clearly showed an original power spectrum pattern exists in the bowel sounds. Moreover, an unknown bowel sound was detected using the standard bowel sound power spectrum as a template. The system we developed clearly detects all bowel sounds perceivable to the human ear.
