Analysis of Cancerous Liver MRI Image using Various Segmentation Methods

Abstract

The precise and early identification of cancerous regions in liver MRI images is paramount for effective diagnosis and treatment planning. This research work focuses on the analysis of edge detection techniques in the context of cancerous liver MRI image segmentation. Leveraging Region of Interest (ROI) selection and the Chan-Vese segmentation method are the main study which aims to improve the accuracy and efficiency of liver cancer localization. In order to improve the accuracy, two phases are involved in this proposed work. The initial phase involves the meticulous selection of ROIs within liver MRI. The ROI-based approach enhances computational efficiency by narrowing down the area of interest and reduces the processing burden. In the second stage,  identify regions that may harbor cancerous lesions, optimizing the subsequent segmentation process. Then, the work investigate the applicability and performance of various edge detection techniques, including classic methods such as Sobel, Canny, and Prewitt. The traditional techniques are essential for extracting meaningful edges and features from the MRI images. The effectiveness of this approach is evaluated concerning their ability to isolate the cancerous areas within the defined ROIs. The Chan-Vese segmentation method, a level set-based active contour approach, is incorporated into the proposed workflow. This method has been recognized mainly for its versatility in handling complex object shapes and variations in intensity, making it a valuable tool for medical image analysis. The comparative analysis is obtained by using Chan-Vese and from traditional edge detection techniques to assess its effectiveness to delineating cancerous liver regions accurately. The evaluation is conducted on a diverse dataset of cancerous liver MRI images, and different performance metrics such as accuracy, sensitivity, feature specification, and the Dice similarity co-efficient are utilized. The research findings of our proposed work highlight the superiority of the Chan-Vese segmentation method when integrated with ROI selection, demonstrating its potential in improving the precision of liver cancer identification. The main goal of this approach is to enhance the accuracy and efficiency of liver cancer diagnosis, contributing to more effective treatment planning and patient care.