MatConvNet-Based Fast Method for Cervical MR Images Classification

Abstract

The deep convolutional neural network (CNN) has been successfully used to obtain high-level representation in various applications of computer vision problems. However, in the field of medical imaging there are not sufficient images available to train a deep CNN. Therefore, we have used a pre-trained deep CNN model for classification of cervical cancer MR images. In this paper, we have proposed MatConvNet-based CNN model to extract features from pre-trained CNN for classification. The vgg-f architecture is deployed to extract the image features. We have evaluated our proposed system with benchmark cervical cancer database obtained from Tumor Cancer Imaging Archive (TCIA). We got the promising result with 98.9% accuracy that is beyond the methods reported in the literature.