Automatic colon polyp classification using Convolutional Neural Network: A Case Study at Basque Country

In this paper, a database generated from the contribution of Urduliz and Biodonostia Hospitals was used, which includes 785 images. The objective is to identify in which category of the Paris classification each image is. 5 ResNet50, ResNet101, Xception, VGG19, and Inception V3 models were used. The model with the best results is ResNet50 with accuracy of 87.1 %, Precision 87.1 %, Recall 87.1 %, F1-Score 87.1 % and Specificity 93 %. In this article, we propose a frame of reference for a set of neural network models, which offers a method for the classification of polypoid or sessile lesions that includes the categorization of Paris.