RCNN with Swallow Swarm Optimization for Liver Disease Detection and Classification

Abstract

One of the most serious medical conditions that can endanger a person's life and health is liver disease. The second-leading cause of death for men and sixth-leading cause of death for women, respectively, is liver cancer. In 2008, liver cancer claimed the lives of almost 750,000 people, killing 960,000 of them. The segmentation and identification of CT images produced by computer tomography has emerged as a major topic in medical image processing. There are few choices for liver segmentation due to the enormous amount of time and resources necessary to train a deep learning model. As part of this research, we created the Region utilizing Convolutional Neural Network, a novel way of extracting the liver from CT scan images (RCNN). The suggested CNN approach, which employs softmax to isolate the liver from the background, contains of three convolutional layers and two entirely associated layers. In the CapsNet and CAL layers, there are class dependencies and an efficient mechanism to connect CAL and subsequent CapsNet processing. Finally, the classification is carried out using the SSO-CSAE model, an approach known as the swallow swarm optimization that is based on the Convolutional Sparse Autoencoder (CSAE) The MICCAI SLiver '07, 3Dircadb01, and LiTS17 benchmark datasets were used to validate the proposed RCNN-SSO approach. When compared to other frameworks, the proposed framework performed well in numerous categories.