TY  - JOUR
AU  - Julius Blöcker, Tom
AU  - Delopoulos, Nikolaos
AU  - Palacios, Miguel A
AU  - Klüter, Sebastian
AU  - Hörner-Rieber, Juliane
AU  - Rippke, Carolin
AU  - Placidi, Lorenzo
AU  - Boldrini, Luca
AU  - Frascino, Vincenzo
AU  - Andratschke, Nicolaus
AU  - Baumgartl, Michael
AU  - Dal Bello, Riccardo
AU  - Marschner, Sebastian N
AU  - Belka, Claus
AU  - Corradini, Stefanie
AU  - Dudas, Denis
AU  - Riboldi, Marco
AU  - Kurz, Christopher
AU  - Landry, Guillaume
TI  - GTV segmentation in MRI guided radiotherapy with promptable foundation models.
JO  - Physics in medicine and biology
VL  - 71
IS  - 1
SN  - 0031-9155
CY  - Bristol
PB  - IOP Publ.
M1  - DKFZ-2025-03070
SP  - 015006
PY  - 2025
N1  - PUBLISHED29 December 2025
AB  - Objective. Magnetic resonance imaging (MRI) guided radiotherapy requires the delineation of gross tumor volumes (GTV) in daily MRI from MRI-linacs. Specialized models have been developed for this task for certain tumors. This study investigated an alternative, using promptable foundation models.Approach. Promptable foundation models were prompted with six different sparse geometric prompt types (points, boxes, 2D masks) to produce GTV segmentation masks, including Segment-anything 2 (SAM2), SAM2 fine-tuned for medical imaging (MedSAM2), and nnInteractive, an nnUnet-based promptable model for medical imaging. A diverse multi-institutional dataset of clinical GTV masks from the abdomen, lung, liver, pancreas, and pelvis sites on MRI scans from MRI-linacs was used to evaluate model outputs using various metrics, including the Dice similarity coefficient (DSC).Main results. The models produced segmentation masks comparable or superior to those from domain-specific models with median DSCs of up to 0.85 (nnInteractive-mask3 prompt). Prompts with more spatial information yielded better results with lower variance, with the effect reduced for nnInteractive and MedSAM2. These produced overall better results (median DSC over all prompt types 0.75 for nnInteractive, 0.70 for MedSAM2, 0.54 for SAM2).Significance. This investigation showed that promptable foundation models can in principle be used for GTV segmentation in MRI across multiple tumor types, although more research is necessary to reduce the variance and improve model performance.
KW  - Magnetic Resonance Imaging
KW  - Humans
KW  - Radiotherapy, Image-Guided: methods
KW  - Tumor Burden
KW  - Image Processing, Computer-Assisted: methods
KW  - Neoplasms: radiotherapy
KW  - Neoplasms: diagnostic imaging
KW  - Neoplasms: pathology
KW  - GTV segmentation (Other)
KW  - MRI-linac (Other)
KW  - MRgRT (Other)
KW  - deep learning (Other)
KW  - foundation models (Other)
KW  - promptable models (Other)
LB  - PUB:(DE-HGF)16
C6  - pmid:41401509
DO  - DOI:10.1088/1361-6560/ae2db9
UR  - https://inrepo02.dkfz.de/record/307471
ER  -