Liver cancer is among the most prevalent and life-threatening cancers worldwide, with hepatocellular carcinoma and liver metastases contributing significantly to cancer related mortality. Accurate segmentation of liver tumors in abdominal computed tomography and magnetic resonance imaging is crucial for tumor prediction, radiation therapy, and treatment monitoring. In recent years, medical image segmentation has advanced rapidly due to deep learning (DL), particularly convolutional neural networks (CNN) and transformer-based architectures. This paper provides a comprehensive review of DL based liver tumor segmentation methods in abdominal imaging. We examine the evolution from traditional encoder decoder frameworks such as U-Net and V-Net to residual, attention based, and hybrid CNN transformer models, including vision transformers. The review also analyzes performance across publicly available benchmarks, including the Medical Segmentation Decathlon, 3D-IRCADb, CHAOS, and the Liver Tumor Segmentation Challenge, using evaluation metrics such as Dice Similarity Coefficient, Hausdorff Distance, and Intersection over Union. Furthermore, we discuss key challenges including tumor heterogeneity, class imbalance, limited annotated data, and domain adaptation across imaging protocols. Emerging directions such as foundation models, self-supervised learning, federated learning, and clinically deployable AI segmentation systems are also highlighted.

Body composition analysis provides essential evidence for evaluating patient health, predicting disease risk, and guiding personalized treatment. Abdominal computed tomography (CT) enables high-resolution visualization of skeletal muscle, adipose tissue, and major organs, making it well-suited for constructing 3D anatomical distribution maps. However, current clinical practice largely depends on manual analysis, which is limited in scope, efficiency, and standardization. This review presents recent deep learning techniques, focusing on methods developed by our group, that aim to automate the generation of anatomical distribution maps from abdominal CT. The key topics include: (1) automated localization and identification of the entire spine beyond single-level detection at the third lumbar vertebra (L3); (2) cross-domain consistency learning to overcome limited annotations in medical imaging; (3) multi-organ segmentation that incorporates inter-slice structural continuity and anatomical relationships; and (4) a comparative study of continual learning and multi-dataset learning for building generalizable models. These approaches enable rapid, standardized generation of anatomical distribution maps across large-scale datasets and diverse patient cohorts, providing a robust technical foundation for future precision medicine applications. In particular, these methods are expected to provide for early cancer detection, pediatric CT analysis, and clinical correlation studies linking anatomical distribution maps with outcomes such as cancer prognosis.