Segmentation and feature extraction to evaluate the stomach dynamic

Abstract

Our work is motivated by a practical application of motion study induced by breathing in the abdomen under the influence of diaphragm pressure. In this article, we present an image processing method to analyze the motion of the stomach between expiration and inspiration. Our approach is based on a preliminary edge detection. To extract useful information, we propose two techniques lying respectively on filling and active contour algorithms. The results of both methods are analysed and compared to those obtained manually. In order to improve the results and to overcome the limitations of the proposed edge detector, an edge closing method based on a polynomial contour fitting is applied.