Review examines how medical imaging foundation models are moved toward clinical use

AI is already widely used for lesion detection, disease classification, and outcome prediction in medical imaging, but most systems remain narrowly designed for one task and depend heavily on expert-labelled data. Their performance can deteriorate when scanners, acquisition protocols, patient populations, or institutions differ from the training environment. Foundation models promise greater reuse through pre-training on large image collections, image-report pairs, clinical text, structured records, and videos. Yet current evidence still comes largely from public benchmarks, retrospective cohorts, and controlled settings, leaving uncertainty about real-world robustness, fairness, and clinical impact. Because of these challenges, deeper research is needed into representative data, prospective validation, workflow integration, and continuous governance.

Researchers from the Institute of Automation, Chinese Academy of Sciences, and the School of Engineering Medicine at Beihang University published (DOI: 10.12290/xhyxzz.2026-0416) the review in July 2026 in the Medical Journal of Peking Union Medical College Hospital. The authors examine how medical imaging foundation models are built, adapted, evaluated, and moved toward clinical use. They also assess progress across radiology, digital pathology, ultrasound, and surgical video, while emphasizing that a central question is not simply whether these models can perform many tasks, but whether they can deliver stable, verifiable value in real-world clinical environments.

The review organizes the field into four overlapping development paths: image-representation pre-training, image-language alignment, integration of multiple clinical data sources, and modelling of dynamic visual sequences. Training data may include computed tomography (CT), magnetic resonance imaging (MRI), X-rays, ultrasound, pathology slides, reports, laboratory measurements, treatment records, endoscopy, and surgical video. The authors explain that data volume alone can be misleading because millions of image patches or video frames may not correspond to an equivalent number of independent patients. Quality control, deduplication, patient-level independence, cross-centre coverage, and reliable pairing between images and text are therefore critical considerations. They also describe adaptation strategies ranging from lightweight task heads to fine-tuning, prompt learning, and instruction tuning. Single-modality examples span cancer subtyping, mutation prediction, survival estimation, and lesion segmentation, while vision-language systems add retrieval and question answering. Crucially, evaluation should extend beyond accuracy. The review proposes examining algorithmic robustness under external data and input disturbances, clinical usefulness through comparisons of clinician-only and clinician-model performance, and workflow outcomes such as reporting time, triage efficiency, repeat examinations, resource use, and patient outcomes. It also stresses that a foundation model may still underperform a task-specific system in a clearly defined clinical setting.

The review argues that the field's next milestone should not be another increase in parameter counts, but evidence that a model can work safely within a defined clinical role. It suggests that a more feasible path may be to use foundation models to assist with limited, reviewable tasks—such as triage, report drafting, interactive segmentation, risk stratification, or structured follow-up—rather than attempting to replace an entire diagnostic process. The review also calls for clear indications, prohibited uses, input-quality requirements, uncertainty signals, human review responsibilities, failure reporting, version tracking, and revalidation after updates.

For clinical deployment, the review outlines a practical path for turning broad technical capability into accountable clinical support. Models should connect reliably with picture archiving and communication systems (PACS), radiology information systems (RIS), and hospital information systems (HIS), while preserving logs of inputs, outputs, clinician edits, warnings, and software versions. Prospective studies should test whether deployment improves decisions, efficiency, or patient outcomes across different centres and patient groups. Governance must also address privacy, consent, secondary data use, copyright, demographic bias, performance drift, and responsibility for errors. The wider implication is that clinical translation will depend on a layered partnership among general foundation models, specialty-specific systems, and human oversight, with each component serving a clearly bounded and auditable role.