Microvascular morphological type recognition using trainable feature extractor

This paper focuses on the problem of feature extraction and the classification task of microvascular morphological type to aid esophageal cancer detection. A specialized convolutional neural network (CNN) is designed to extract hierarchical features and Support Vector Machines (SVMs) are introduced to enhance the generalization ability of classifiers. Experiments are conducted on the NBI-ME dataset, achieving a recognition rate of 88.19% on patch level. The results show that the CNN-SVM model beats models of traditional features with SVM as well as the original CNN with softmax. The synthesis results indicate this system is able to assist clinical diagnose to a certain extent.