Proposing method to Increase the detection accuracy of stomach cancer based on colour and lint features of tongue using CNN and SVM

TL;DR

This paper presents a novel method combining deep CNN and SVM to improve early gastric cancer detection accuracy from tongue images, achieving 91% accuracy and outperforming existing methods.

Contribution

It introduces a new approach using deep RCNN for tongue region segmentation and combines CNN with SVM for enhanced gastric cancer detection accuracy.

Findings

DenseNet achieved the highest accuracy among tested architectures.

The proposed method reached 91% detection accuracy.

The approach outperforms existing state-of-the-art methods.

Abstract

Today, gastric cancer is one of the diseases which affected many people's life. Early detection and accuracy are the main and crucial challenges in finding this kind of cancer. In this paper, a method to increase the accuracy of the diagnosis of detecting cancer using lint and colour features of tongue based on deep convolutional neural networks and support vector machine is proposed. In the proposed method, the region of tongue is first separated from the face image by {deep RCNN} \color{black} Recursive Convolutional Neural Network (R-CNN) \color{black}. After the necessary preprocessing, the images to the convolutional neural network are provided and the training and test operations are triggered. The results show that the proposed method is correctly able to identify the area of the tongue as well as the patient's person from the non-patient. Based on experiments, the DenseNet network has the highest accuracy compared to other deep architectures. The experimental results show that the accuracy of this network for gastric cancer detection reaches 91% which shows the superiority of method in comparison to the state-of-the-art methods.