Autofluorescence Imaging Video-Endoscopy System for Diagnosis of Superficial Gastric Neoplasia

Abstract

An autofluorescence imaging video-endoscopy system (AFI) produces real-time pseudocolor images from computed detection of autofluorescence emitted by endogenous fluorophores in the mucosa. In the AFI images of the gastric body, the fundic mucosa appears purple, whereas atrophic mucosa appear bright green. Gastric tumors appear purple or green in the AFI images according to their morphology, i.e., elevated or depressed, respectively. Therefore, the color patterns of gastric tumors are classified into four types: purple tumors in a green background, purple tumors in a purple background, green tumors in a green background, and green tumors in a purple background. Purple tumors in a green background and green tumors in a purple background are readily distinguished by their color. In contrast, purple tumors in a purple background are difficult to define by color. Green tumors in a green background appear similar in color to the surrounding mucosa, but the tumor extension can be determined by their purple rim. Diagnostic accuracy of AFI for tumor extension was better (68%) than that by white light endoscopy (36%), but was not as good as chromoendoscopy (91%). The low accuracy rate was mainly caused by interference of an ulceration or scar, while the AFI diagnosed flat tumor extension more accurately than did white light images. Because the AFI could visualize flat or isochromatic tumor extension compared with white light endoscopy, it detected more multiple neoplasia in patients who underwent endoscopic treatment.