Evaluation of automated organ segmentation for total-body PET-CT

Abstract

The ability to diagnose rapidly and accurately and treat patients is substantially facilitated by medical images. Radiologists' visual assessment of medical images is crucial to their study. Segmenting images for diagnostic purposes is a crucial step in the medical imaging process. The purpose of medical image segmentation is to locate and isolate ‘Regions of Interest’ (ROI) within a medical image. Several medical uses rely on this procedure, including diagnosis, patient management, and medical study. Medical image segmentation has applications beyond just diagnosis and treatment planning. Quantitative information from medical images can be extracted by image segmentation and employed in the research of new diagnostic and treatment procedures. In addition, image segmentation is a critical procedure in several programs for image processing, including image fusion and registration. In order to construct a single, high-resolution, high-contrast image of an item or organ from several images, a process called "image registration" is used. A more complete picture of the patient's anatomy can be obtained through image fusion, which entails integrating numerous images from different modalities such as computed tomography (CT) and Magnetic resonance imaging (MRI). Once images are obtained using imaging technologies, they go through post-processing procedures before being analyzed. One of the primary and essential steps in post-processing is image segmentation, which involves dividing the images into parts and utilizing only the relevant sections for analysis. This project explores various imaging technologies and tools that can be utilized for image segmentation. Many open-source imaging tools are available for segmenting medical images across various applications. The objective of this study is to use the Jaccard index to evaluate the degree of similarity between the segmentations produced by various medical image visualization and analysis programs.