Detection of Liver Cancer through Computed Tomography Images using Deep Convolutional Neural Networks

Liver cancer is the fifth most common type of tumor in men and the ninth most common type of tumor in women. After taking a sample of liver tissue, imaging tests like computed tomography (CT), ultrasound, and magnetic resonance imaging (MRI), can be used to diagnose the liver tumor. In recent studies, accurate detection of liver cancer with minimum computational time and computational complexity is a major issue remained the challenge. This research proposes a framework to segment the cancerous area through CT scan images using entropy thresholding technique. Additionally, it uses two CNN models, U-Net and Google-Net, for the classification of liver cancer. The proposed method employs the 3D-IRCADb01 dataset, which consists of CT slices of liver tumor patients. The U-Net performed better than other networks with 98.5% accuracy, 0.83 DSC, 99.5% recall, and 98.75% F1. Biotechnology uses this method for an early and accurate diagnosis of liver tumors that is likely to save many lives. Proposed method outperformed than existing state-of-art methods and is suitable for clinical applications to assist doctors in diagnosing liver cancer.