Artificial intelligence technology for automatic segmentation of lymphatic areas in radiotherapy for pelvic tumors

The clinical treatments of pelvic cancers, such as cervical cancer, adenocarcinoma, and prostate cancer, include not only surgical resection but also radiotherapy. Radiotherapy uses radiation to target and control the development of the disease. Before radiotherapy, precise marking and delineation are essential for radiation dose planning. The delineated areas include the lesion site and clinically determined potential areas, such as lymph nodes and surrounding organs at risk (OAR), including the bladder, uterus, large intestine, and small intestine. Currently, the clinical labeling of these lesions and OAR areas is done manually by experienced radiation oncologists and medical physicists. The pelvic cavity's anatomical structure is complex, and the lymph node area in the pelvic cavity is recognized as a high-risk area, thus included in the irradiation area. However, lymph nodes on CT images have fewer clear borders than other OAR areas, making it difficult to delineate and increasing variability between different markers. This study employs artificial intelligence algorithms to automatically delineate clinical target volumes (CTV), especially pelvic lymph nodes, and attempts to incorporate anatomical information to facilitate more accurate annotation. This approach aims to reduce the burden of clinical treatment planning and improve precision in radiotherapy.