Automated diagnosis of Barrett's esophagus with endoscopic images

In this paper, we describe current progress on the development of a Computer Assisted Diagnosis System (CAD) for the classification of Barrett's esophagus and associated neoplasia. Barrett's esophagus is a condition in which normal squamous mucosa is replaced by columnar epithelium, which is similar to the lining of the intestine. Barrett's esophagus as a known precancerous condition leading to esophageal cancer. Diagnosis is performed via histological analysis of tissue located during endoscopic examination. We compare four different automated classification tools (SVM, KNN, and Boosting) operating on three different imaging modalities (white light, narrow-band, and acetic acid chromoendoscopy) for lesion classification. Preliminary results suggest that narrow band imaging is more effective than either of the other two modalities for disease assessment.