Increasing the accuracy in the diagnosis of stomach cancer based on color and lint features of tongue. (August 2021)

• Record Type: Journal Article
• Authors: Gholami, Elham; Kamel Tabbakh, Seyed Reza; kheirabadi, Maryam
• Is Part Of: Biomedical signal processing and control. Volume 69(2021)
• Publication Date: 2021-08
• Subjects: Gastric cancer -- Support vector machine -- Convolutional neural networks -- Dense convolutional network; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2021.102782
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: A method to increase the accuracy of the diagnosis of detection cancer using lint and color 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 Recursive Convolutional Neural Network (R-CNN). 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 identifying the area of the tongue as well as the patient's person from the non-patient. Abstract: Gastric cancer is the second most common cancer in the world. Due to the time-consuming diagnosis of the disease, it is essential to use new methods (e.g., computer science) for early diagnosis. Among various computer-based detection methods, artificial intelligence algorithms have attracted great attention today. The present study aimed to increase the accuracy of gastric cancer diagnosis by using a combination of deep neural network, support vector machine, and deep convolutional neural network (CNN) based on the surface and color features of the tongue. The proposed method was evaluated in seven CNN architectures. According to the results, using the DenseNet architecture in the proposed method had a higher accuracy compared to the other architectures, and 91 % accuracy was observed in the diagnosis of gastric cancer.