A computer-aided lesion diagnose method based on gastroscopeimage

It is time-consuming and exhausting for clinicians to make gastrointestinal disease diagnosis by gastroscope images (GI) with naked eye. However, method use GI as data source to make computer aided diagnosis does not exist at present. In this paper, we discussed a computer aided method for GI, which can be divided into two parts: First, take samples in images which were labeled by clinicians, and extract RG, OPPO, HUE color histogram, and LBP texture in these samples, then use SVM classifier to make classification, the classification accuracy of the above characteristics are given respectively. Second, the detecting images should be divided into little patch, for each patch use features mentioned above and SVM classifier to make classification, then, get the position of the diseased areas. Finally, through experiments shows the effectiveness of the method, and give its diagnostic accuracy rate.