Transformer models show mixed robustness in radiation therapy planning

Researchers have developed a transformer-based pipeline for predicting fluence maps in radiation therapy, aiming to speed up treatment planning. Their study evaluated the model's robustness against various clinically realistic perturbations, including geometric shifts, noise, and domain changes. The findings indicated that while the model degrades smoothly under moderate disturbances, it fails sharply with severe rotations and noise, with hierarchical transformers showing better resilience. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT This research explores the robustness of AI models in medical applications, highlighting potential failure modes and the need for physics-informed evaluation in radiation therapy.

RANK_REASON This is a research paper detailing a new method and its evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• SwinUNETR

• paper
• other