Deep learning and Colon Capsule Endoscopy: automatic detection of blood and colonic mucosal lesions using a convolutional neural network 

Colon capsule endoscopy is progressively becoming an accepted non-invasive alternative to conventional colonoscopy. Nevertheless, this attribute is hampered by its limitations, most importantly the time required for reading each exam. Our group developed a convolutional neural network for the automatic detection and differentiation of colonic blood and multiple colonic lesions. describe the development of a deep learning algorithm for automatic detection of colonic blood and hematic residues in colon capsule endoscopy. Our tool had a mean sensitivity of 96% and specificity of 98%. This article is pioneer for the further development of artificial intelligence algorithms to mitigate the current limitations of colon capsule endoscopy and improve its diagnostic efficiency.

The article can be found here.

In Press // February 2022