Related Experiment Video
Updated: Jul 9, 2026
07:14
Virtual Agent for Real-Time Motivational Interviewing by Integrating Adaptive Nonverbal Behavior and Language Models
Published on: December 23, 2025
Human-like AI-based auto-field-in-field whole-brain radiotherapy treatment planning with conversation large language
Adnan Jafar1, An Qin1, Gavin Atkins1
1Department of Radiation Oncology and Molecular Radiation Sciences, Johns Hopkins University, Baltimore, Maryland, USA.
Journal of Applied Clinical Medical Physics
|July 8, 2026
Summary
This study introduces an automated system for whole-brain radiotherapy planning using deep learning and large language models. The AI-driven approach significantly speeds up treatment planning while maintaining high clinical quality and patient safety.
Area of Science:
- Radiation Oncology
- Medical Physics
- Artificial Intelligence in Medicine
Background:
- Whole-brain radiotherapy (WBRT) is a standard treatment, but its planning is labor-intensive.
- Current automated planning tools require manual optimization and refinement.
- Patient-specific adjustments are crucial for optimal WBRT plan generation.
Purpose of the Study:
- To develop an automated WBRT planning pipeline using deep learning (DL) and large language models (LLMs).
- To integrate a DL model for predicting patient-specific hyperparameters for automated planning.
- To implement an LLM-based interface for interactive refinement of WBRT plans.
Main Methods:
- A DL Hyperparameter Prediction model was trained on 55 WBRT cases using geometric features.
- Automated Field-in-Field (Auto-FiF) settings were predicted for the RayStation treatment planning system.
- An LLM (GPT-4o) processed voice feedback for plan adjustments via a conversational interface, using Whisper for transcription.
- Plan quality was assessed in 15 independent cases using clinical metrics and expert review.
Main Results:
- 14 out of 15 DL-generated WBRT plans were clinically acceptable.
- DL-generated plans showed no significant difference in dose metrics compared to manual plans.
- Automated planning reduced workflow time to approximately 7 minutes from 15 minutes for manual planning.
- The conversational module effectively improved dose conformity and reduced hotspots in suboptimal plans.
Conclusions:
- The integrated DL and LLM system enhances WBRT planning efficiency.
- The approach maintains clinically acceptable plan quality.
- This study demonstrates the feasibility of AI-driven WBRT planning for streamlined, high-quality treatments.
