Automatic Segmentation and Categorization of Liver Tumors in CT Scans: A Bibliometric Analysis

Liver tumors rank as the third deadliest cancer globally and the sixth most prevalent disease worldwide. They primarily afflict individuals who frequently consume tobacco or alcohol. Approximately 75-85 percent of primary liver cancer cases are attributed to these factors. However, manually diagnosing liver tumors presents significant challenges due to tumor heterogeneity, varying shapes and sizes, imaging artifacts, and limited annotated data. The critical task of segmenting liver tumors in medical images greatly impacts diagnosis and treatment planning. Numerous techniques and frameworks have been formulated for the early identification of tumors, improving accuracy, and aiding doctors in understanding tumor characteristics such as size and volume. Nevertheless, this process is error-prone and cumbersome. To address these challenges, a proper model is needed to identify the tumor and used for real-time image processing. This paper motivates to develop a system and assists doctors in swiftly identifying and segmenting tumors from images, providing an approximation of tumor size and stage for more precise treatment. Manual diagnosis is challenging, but advanced DL methods like TransUNet help with early detection and tumor segmentation, assisting doctors in providing timely and accurate treatment.