Domain-aligned self-supervision: CT-specialized DINOv3 enhances PET/CT tumor segmentation
Open Access
- 23 October 2026
- journal article
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 71 (20) , 205022
Abstract
Objective. This study investigates how domain-aligned self-supervised pretraining influences downstream PET/CT tumor segmentation, with a focus on disentangling the roles of domain proximity and pretraining dataset scale. Approach. We compared nnU-Net segmentation models initialized from random weights or from DINOv3 encoders pretrained on natural images, x-ray mammography, CT, or PET, using datasets of different sizes. To disentangle the effects of pretraining domain and dataset size, CT-DINOv3 encoders were pretrained using different numbers of 2D CT slices, including subsets matched in size to the pretraining datasets from other domains. Models were evaluated on two PET/CT benchmarks, autoPET (whole-body, FDG/prostate-specific membrane antigen (PSMA)) and HECKTOR (head-and-neck primary tumors and metastatic lymph nodes), using five-fold cross-validation with frozen encoders. Main results. CT-pretrained models achieved the highest Dice scores for lymph nodes, FDG lesions, and PSMA lesions. For primary tumors, PET pretraining produced the highest observed Dice score, followed by CT pretraining, suggesting the potential benefit of in-domain representations when PET data are sufficient. Notably, PET pretraining exceeded size-matched CT pretraining for PSMA but remained below models pretrained on larger CT datasets for both PSMA and FDG targets. In contrast, natural-image pretraining did not outperform from-scratch training despite its much larger pretraining scale, indicating that scale alone is insufficient when domain proximity is low. Across comparisons, strong transfer reflected the combination of domain proximity and sufficient pretraining scale, especially for CT-based models with broad data availability. Significance. These results demonstrate that both domain proximity and dataset scale jointly determine the effectiveness of self-supervised pretraining. While higher domain alignment improves performance at comparable scale, sufficient data is required to realize this benefit. CT-based pretraining provides a favorable balance between domain relevance and data availability, leading to robust performance across tasks. This study provides practical guidance for selecting pretraining strategies in PET/CT segmentation.Keywords
Funding Information
- National Institute of Biomedical Imaging and Bioengineering (R01EB032680)
- National Institute of Dental and Craniofacial Research (R01DE033512)
- National Heart, Lung, and Blood Institute (R01HL163570)
- National Cancer Institute (R01CA272991)
This publication has 28 references indexed in Scilit:
- Necessity and impact of specialization of large foundation model for medical segmentation tasksMedical Physics, 2024
- Positron Emission Tomography (PET)/Computed Tomography (CT) Imaging in Radiation Therapy Treatment Planning: A Review of PET Imaging Tracers and Methods to Incorporate PET/CTAdvances in Radiation Oncology, 2023
- A General Web-Based Platform for Automatic Delineation of Head and Neck Gross Tumor Volumes in PET/CT Images2023
- Simplicity Is All You Need: Out-of-the-Box nnUNet Followed by Binary-Weighted Radiomic Model for Segmentation and Outcome Prediction in Head and Neck PET/CT2023
- Overview of the HECKTOR Challenge at MICCAI 2022: Automatic Head and Neck Tumor Segmentation and Outcome Prediction in PET/CT2023
- Multi-scale Fusion Methodologies for Head and Neck Tumor Segmentation2023
- Deep Learning and Machine Learning Techniques for Automated PET/CT Segmentation and Survival Prediction in Head and Neck Cancer2023
- Transfer learning in medical image segmentation: New insights from analysis of the dynamics of model parameters and learned representationsArtificial Intelligence in Medicine, 2021
- FDG PET/CT: EANM procedure guidelines for tumour imaging: version 2.0European Journal of Nuclear Medicine and Molecular Imaging, 2014
- Combined 18F-FDG-PET/CT Imaging in Radiotherapy Target Delineation for Head-and-Neck CancerInternational Journal of Radiation Oncology - Biology - Physics (IJROBP), 2009