Sparse Bayesian Feature Selection Applied to Intestinal Motility Analysis

In this paper we propose an innovative automatic system for diagnosing severe intestinal motility dysfunctions based on a joint feature selection and classifier learning. A novel acquisition technology, the Wireless Capsule Video Endoscopy, is used as source of data and visual information is exploited to characterize intestinal motility dysfunctions. In particular, the method automatically selects the most relevant visual features among a set of them defined by medical experts and, simultaneously, carries out the classification training. Thus, the dimension of the input data is reduced without any accuracy loss. To assess the accuracy of the proposed method we use the results obtained with the currently applied invasive diagnosis test, the manometry.