Contourlet based feature extraction and classification for Wireless Capsule Endoscopic images

Wireless Capsule Endoscopy (WCE) is a late-model non-invasive device to detect abnormalities in small intestine. The traditional diagnostic method that only depending on clinicians' naked eyes is time-consuming and labor-intensive. It is necessary to develop a computer-aided system to alleviate the burden of clinicians. In this paper, a new color-texture feature extraction method is proposed for the classification of normal and abnormal WCE tissue images. The Contourlet Transform is introduced and used for each color channel of each WCE image in HSV color space. Finally, we construct a 288-dimensions feature vector by calculating the 3-order color moments for each baseband generated by using the Contourlet Transform. Real experiments using different classifiers in various color spaces are implemented to evaluate the performance of the proposed method.