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

doi:Recent advances in AI foundation models have significant potential for lightening the clinical workload by mimicking the comprehensive and multi-faceted approaches used by medical professionals. In the field of radiation oncology, the integration of multiple modalities holds great importance, so the opportunity of foundational model is abundant. Inspired by this, here we present RO-LMM, a multi-purpose, comprehensive large multimodal model (LMM) tailored for the field of radiation oncology. This model effectively manages a series of tasks within the clinical workflow, including clinical context summarization, radiotherapy strategy suggestion, and plan-guided target volume segmentation by leveraging the capabilities of LMM. In particular, to perform consecutive clinical tasks without error accumulation, we present a novel Consistency Embedding Fine-Tuning (CEFTune) technique, which boosts LMM's robustness to noisy inputs while preserving the consistency of handling clean inputs. We further extend this concept to LMM-driven segmentation framework, leading to a novel Consistency Embedding Segmentation (CESEG) techniques. Experimental results including multi-center validation confirm that our RO-LMM with CEFTune and CESEG results in promising performance for multiple clinical tasks with generalization capabilities.

YUHSpace

BROWSE

0 1

Cited 0 times in

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

DC Field	Value	Language
dc.contributor.author	김용배	-
dc.contributor.author	김진성	-
dc.contributor.author	박상준	-
dc.contributor.author	변화경	-
dc.date.accessioned	2025-10-15T01:34:32Z	-
dc.date.available	2025-10-15T01:34:32Z	-
dc.date.issued	2025-10	-
dc.identifier.issn	1361-8415	-
dc.identifier.uri	https://ir.ymlib.yonsei.ac.kr/handle/22282913/207411	-
dc.description.abstract	Clinical report; Large multimodal model; Radiation oncology; Radiotherapy target volume; Segmentation	-
dc.description.statementOfResponsibility	https://www.sciencedirect.com/science/article/pii/S1361841525001938	-
dc.language	English	-
dc.publisher	Elsevier	-
dc.relation.isPartOf	MEDICAL IMAGE ANALYSIS	-
dc.rights	CC BY-NC-ND 2.0 KR	-
dc.subject.MESH	1	-
dc.title	End-to-end breast cancer radiotherapy planning via LMMs with consistency embedding	-
dc.type	Article	-
dc.contributor.college	College of Medicine (의과대학)	-
dc.contributor.department	Dept. of Radiation Oncology (방사선종양학교실)	-
dc.contributor.googleauthor	Kwanyoung Kim	-
dc.contributor.googleauthor	Yujin Oh	-
dc.contributor.googleauthor	Sangjoon Park	-
dc.contributor.googleauthor	Hwa Kyung Byun	-
dc.contributor.googleauthor	Joongyo Lee	-
dc.contributor.googleauthor	Jin Sung Kim	-
dc.contributor.googleauthor	Yong Bae Kim	-
dc.contributor.googleauthor	Jong Chul Ye	-
dc.identifier.doi	Recent advances in AI foundation models have significant potential for lightening the clinical workload by mimicking the comprehensive and multi-faceted approaches used by medical professionals. In the field of radiation oncology, the integration of multiple modalities holds great importance, so the opportunity of foundational model is abundant. Inspired by this, here we present RO-LMM, a multi-purpose, comprehensive large multimodal model (LMM) tailored for the field of radiation oncology. This model effectively manages a series of tasks within the clinical workflow, including clinical context summarization, radiotherapy strategy suggestion, and plan-guided target volume segmentation by leveraging the capabilities of LMM. In particular, to perform consecutive clinical tasks without error accumulation, we present a novel Consistency Embedding Fine-Tuning (CEFTune) technique, which boosts LMM's robustness to noisy inputs while preserving the consistency of handling clean inputs. We further extend this concept to LMM-driven segmentation framework, leading to a novel Consistency Embedding Segmentation (CESEG) techniques. Experimental results including multi-center validation confirm that our RO-LMM with CEFTune and CESEG results in promising performance for multiple clinical tasks with generalization capabilities.	-
dc.contributor.localId	A00744	-
dc.contributor.localId	A04548	-
dc.contributor.localId	A06513	-
dc.contributor.localId	A05136	-
dc.relation.journalcode	J02201	-
dc.identifier.eissn	1361-8423	-
dc.identifier.pmid	40516483	-
dc.subject.keyword	Algorithms	-
dc.subject.keyword	Breast Neoplasms* / diagnostic imaging	-
dc.subject.keyword	Breast Neoplasms* / radiotherapy	-
dc.subject.keyword	Female	-
dc.subject.keyword	Humans	-
dc.subject.keyword	Radiotherapy Planning, Computer-Assisted* / methods	-
dc.contributor.alternativeName	Kim, Yong Bae	-
dc.contributor.affiliatedAuthor	김용배	-
dc.contributor.affiliatedAuthor	김진성	-
dc.contributor.affiliatedAuthor	박상준	-
dc.contributor.affiliatedAuthor	변화경	-
dc.citation.volume	105	-
dc.citation.startPage	103646	-
dc.identifier.bibliographicCitation	MEDICAL IMAGE ANALYSIS, Vol.105 : 103646, 2025-10	-
dc.identifier.articleno	10.1016/j.media.2025.103646	-

Appears in Collections:: 1. College of Medicine (의과대학) > Dept. of Radiation Oncology (방사선종양학교실) > 1. Journal Papers

Show simple item record Find it @ YMLIB

License

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

YUHSpace

BROWSE

Browse

Links