Towards an Interpretation of Intestinal Motility Using Capsule Endoscopy Image Sequences

Human intestinal motility is presented by the propagation of peristaltic waves with their frequencies gradually decreasing along the length of the small bowel. This paper describes a heuristic method, which can be used towards interpreting intestinal motility through recognizing their frequency characteristics from capsule endoscopy image sequences. First, image features that reflect peristaltic activities are extracted to build a functional signal. Then, a Multi-Resolution Analysis technique in the wavelet domain is used to decompose the functional signal taking into account the non-stationary nature of intestinal motility. For peristaltic waveform recognition, the method relies on the principle of peak detections from the decomposed signals. Each waveform is detected when it exceeds a baseline level. The frequency characteristics are interpreted through analysis of the waveform appearance and their velocity propagation. Three healthy sequences were tested in experiments. The estimated trends of the peristaltic wave propagation from the experimental results show a frequency gradient, which follows the well-recognized characteristics of intestinal motility propagation. Therefore, this study is the first demonstration of a detailed interpretation of intestinal motility, and we suggest that further research focuses on intestinal motility dysfunctions.