Glossary
Radiomics
Also called: quantitative imaging features
Extracting large numbers of quantitative features from medical images to use as inputs to statistical or machine learning models.
Radiomics computes texture, shape, and intensity features from a defined region and uses them for prediction. It predates deep learning as an approach and remains common in oncology imaging research.
Where This Gets Misread
Radiomic features are highly sensitive to acquisition and reconstruction parameters, and reproducibility across scanners has been a persistent problem. A large published literature reports findings that have not replicated, and feature counts in the hundreds against small patient cohorts is a recipe for spurious association.