From Anatomy to Radiology to Imaging Science: How Images Become Biomarkers and Therapeutic Evidence

Anatomy provides concepts by which form, spatial relationships, and function are brought into context, while radiology extends these concepts into the living. Imaging science asks how biological information is encoded in an imaging signal, converted into a measurement, validated, and used to support therapeutic development and decision-making.

This invited review article follows a pathway from disease biology and image acquisition to quantitative phenotypes, imaging biomarkers, endpoints, and surrogates for regulatory drug approval. What can be measured from an image and how meaningful this is depends on factors such as modality, contrast mechanisms, positioning/loading/motion, reconstruction parameters, and standardization. A quantitative output only becomes a biomarker within a prespecified context of use, and a surrogate after appropriate qualification. The successful development of bone mineral density as a surrogate endpoint for fracture risk illustrates that the association with a clinical outcome is insufficient, whilst scalability and predictive validity of treatment benefits across different drug trials and mechanisms are mandatory. Artificial intelligence can accelerate image acquisition, analysis, and discovery, but does not remove the need for transparency and validation.

Imaging science is not just anatomy displayed from an acquisition. It is a rigorous interdisciplinary science and "art" at the interface of anatomy, physics, radiology, informatics, metrology, statistics, and trials. Its task is to establish when an image-derived signal can be trusted as evidence. With this in mind, imaging biomarkers and surrogate endpoints represent powerful, scalable tools to support clinical programs of modern therapeutic development.