Features for Discriminating Helicobacter Pylori Infection from Gastric X-ray Images

This paper presents a method for extracting effective image features from double contrast X-ray images of stomach to discriminate Helicobacter pylori infection with the images. In the proposed method, after the area for diagnosis is determined, the proposed features are extracted from the area based on characteristics of the images with the infection. In the images, a pattern of folds is shown in the area and diagnosticians diagnose the infection or a normal case reading the pattern. The features are designed according to the standard for reading the fold patterns of the infection. In addition to quantitative evaluation for the infection, the proposed method discriminates the images into normal and infection cases using a learning machine regarding the proposed features as variables. Experimental results obtained by applying the proposed method to the X-ray images have shown effectiveness of the proposed features.