Mar 2026· Artificial Intelligence Review· 0 citations· 195 references
Computer Science
TL;DR
FM adaptation is a context-dependent design space rather than a hierarchy of preferred strategies rather than a hierarchy of preferred strategies; outcomes varied by task, data regime, comparator, and evaluation setting.
Abstract
Foundation models, pretrained on diverse data, objectives, or modalities for reuse across downstream tasks, are increasingly applied to medical image analysis. Their performance depends not only on pretraining but also on how they are adapted to specific tasks, data regimes, and deployment settings. Whereas previous reviews have considered adaptation within broader syntheses, this review focuses specifically on post-pretraining interventions applied to existing FMs. Using a structured narrative review with a systematic, auditable workflow, we identified 719 eligible reports representing 705 independent studies. Of these, 473 provided controlled comparative evidence: 322 formally published studies formed the primary synthesis and 151 preprint-only studies a sensitivity layer. Adaptation was organized into five non-mutually exclusive families: parameter-level, representation-level, objective-level, data-centric, and architectural/sequence-level adaptation, with model use and update regime analyzed separately. Controlled comparisons showed no consistent ranking among frozen, parameter-efficient, partial, or full updating, or across broader adaptation families; outcomes varied by task, data regime, comparator, and evaluation setting. Clinical-readiness evidence remained limited: 173/705 studies (24.5%) had confirmed independent external validation, 10 (1.4%) reported calibration or uncertainty evidence, and 97/322 formally published comparative studies (30.1%) included a qualifying externally validated arm. Overall, FM adaptation is a context-dependent design space rather than a hierarchy of preferred strategies.
This review provides a structured reference to guide MVFMs from technical validation to clinically effective, reliable, and ethically sound deployment and introduces the Data-Technology-Translation Triad as a unifying conceptual framework to structure the progress, key factors, and challenges of MVFMs.
Neng-Jian Lin, Yi Ding, Zhen Qin et al.· Computerized Medical Imaging...· 0 citations
Artificial intelligence (AI) is increasingly used in medical imaging, with research spanning task-specific classification, detection, and segmentation, representation learning, image reconstruction, generative models, foundation models, and vision–language models (VLMs). This review synthesizes 74 core publications and...
Foundation models pretrained on large natural-image collections are increasingly adapted for medical image segmentation, in the expectation that they will replace models trained for each task. Most comparisons, however, use data from the training distribution and a single task-specific baseline whose configuration is r...
Vision-language foundation models (VLFMs) for medical image understanding have advanced rapidly, yet evidence supporting clinical deployment remains fragmented. This narrative review synthesizes architectures, training paradigms, benchmarks, public datasets, and translation barriers. Drawing on more than 50 original st...
Shahid Mehmood, Farman Matloob Khan, Abeer A. K. Alharbi et al.· Frontiers in Medicine· 0 citations
Medical image segmentation is essential for computer-aided diagnosis and precision medicine, yet conventional supervised models depend heavily on high-quality annotations and often generalize poorly across domains. The Segment Anything Model (SAM) family offers a transferable, interactive, and extensible foundation for...
Qing-Ling Xia, Xue Sun, Bin Jiang et al.· Computerized Medical Imaging...· 0 citations
Deep learning supports many tasks in medical imaging, but most systems assume a fixed data source and a single training event. In clinical practice, new scanners arrive, protocols change, patient populations shift, and new diagnostic questions appear. A frozen model loses accuracy as the data shift, while a model retra...
Saqib Qamar, Homod bin Hamed bin Musaed Al-Aloye, Furqan Alam· Bioengineering· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.