A method for recognizing changes in stomach mucosal microstructure by video endoscopy

Abstract

This paper proposes the method for analysis and subsequent recognition of stomach mucosal changes by video endoscopy, including real time mode. The method is based on the parallel use of image analysis algorithms and a neural network. The model for the quantitative estimate of changes in the mucosal microstructure is proposed. The paper performs the correlation analysis of the results from the estimate of mucosal changes that was obtained by the image analysis algorithms and using the neural network. The accuracy of neoplasm recognition in endoscopic images is estimated for different combinations of algorithms and descriptors.