Análisis de la motilidad intestinal utilizando Convolutional Deep Neural Network y la cápsula endoscópica

Abstract

Capsule endoscopy is a diagnostic technique that opens a wide field of research in the clinical area. Through the capsule, videos of the entire digestive tract can be obtained, as it is ingested by the patient. There are many projects related to the data extraction of these videos, one of them, a Convolutional Deep Neural Network which is able to classify images of the small intestine in six different classes. Nevertheless, since the v ́ıdeo is obtained until it's able to be processed by the Convolutional Deep Neural Network, a set of operations are needed. In the first part of this document, an approach to get this work done is proposed. In the second part, the results of the neural network are used to create an application that tries to find sequences in videos and segment the videos based on that sequences.