Deep Learning Approach for Liver Tumor Diagnosis

Liver cancer is a significant issue in Egypt, as indicated by the most recent statistics from the World Health Organisation in 2020, where liver cancer-related deaths reached 4.57 percent of the total mortality rate. Egypt is ranked second worldwide. The immediate detection and correct diagnosis of liver cancer are essential for effective treatment and improved patient outcomes. In recent years, deep learning techniques have been utilized to enhance the accuracy of liver cancer detection and diagnosis. This paper introduces a significant effort to collect and process a real dataset. The liver tumor dataset was gathered from Egypt’s Ain Shams University, Specialized Hospital (ASUSH). The proposed liver dataset includes 223 patients with both healthy and diseased conditions. The dataset comprises a total of 13,287 computed tomography (CT) images. Images were obtained from the PACS system and converted from the Digital Imaging and Communications in Medicine (DICOM) to Tiff image format. The paper also utilizes and compares the results of various convolution neural networks (CNN)-based deep learning models on the collected data.