Document Type : Research

Authors

• Pegah Davrpanah ¹
• Somayeh Saraf Esmaili ²

¹ Ph.D. Student in Biomedical Engineering, Department of Biomedical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran.

² Assistant Professor, Department of Biomedical Engineering, Garmsar Branch, Islamic Azad University, Garmsar, Iran.

Abstract

Liver cancer is the fifth most common cancer in men and the seventh most common cancer in women, and is the third leading cause of cancer-related deaths worldwide. In general, the disease is less common in women, and in most regions of the world, the rate of liver cancer in men is two to three times higher than in women, which is probably due to the higher prevalence of risk factors in men and gender differences. In this regard, the aim of this research was to detect liver cancer from CT scan images using convolutional neural network and support vector machine. In this research, a new lightweight CNN neural network with seven layers and only one conventional layer is proposed for segmented liver classification. This proposed model was used in two different ways. The first path used deep learning classification and achieved 83.7% and 95.9% accuracy. Meanwhile, the second route used automatically extracted features together with a support vector machine (SVM) classifier and achieved 95.9% and 97.9% accuracy. The proposed network is lightweight, fast, reliable and accurate. This approach can be used by an oncologist, making detection a simple task. In addition, the proposed network achieves high accuracy without adjusting images, which reduces time and cost.

Keywords

• Detection of liver cancer
• convolutional neural network
• support vector machine