Intelligent Support for Radiotherapy: A Review of Clinical Applications for Large Language Models

Area of Science:

• Artificial Intelligence in Oncology
• Medical Imaging and Radiation Therapy

Background:

• Radiation therapy (RT) faces challenges with manual target delineation variability and data integration.
• Large language models (LLMs) offer advanced semantic understanding and cross-modal fusion for RT.

Purpose of the Study:

• To provide a comprehensive review of LLM applications across the radiation therapy workflow.
• To explore the potential and limitations of LLMs in advancing RT.

Main Methods:

• Narrative review of current literature on LLM applications in radiation oncology.
• Analysis of LLM utility in various RT domains, including delineation, planning, and decision support.

Main Results:

• LLMs show utility in automated target delineation, dose prediction, treatment planning, and clinical decision support.
• Identified limitations include model hallucinations, generalizability issues, and data privacy concerns.

Conclusions:

• LLMs have transformative potential for radiation oncology, enhancing accuracy and efficiency.
• Future work must focus on technical refinement, standardized benchmarks, and ethical frameworks for clinical integration.