End-to-end breast cancer radiotherapy planning via LMMs with consistency embedding

Kwanyoung Kim, Yujin Oh, Sangjoon Park, Hwa Kyung Byun, Joongyo Lee, Jin Sung Kim, Yong Bae Kim, Jong Chul Ye

Abstract: Highlights•We propose an LMM framework to support core radiation oncology workflows.•Our model enables report summarization, plan suggestion, and volume segmentation.•We introduce novel noise augmentation and consistency training techniques.•Multi-centric evaluation validates our model’s potential for real clinical settings.