Radiomics: Images Are More than Pictures, They Are Data¶
Why this mattered¶
Gillies, Kinahan, and Hricak helped crystallize a shift in medical imaging from qualitative interpretation toward quantitative, computable phenotyping. The key claim was not simply that algorithms could assist radiologists, but that standard-of-care CT, MRI, and PET images already contained large amounts of latent biological and clinical information. By framing images as “mineable data,” the paper gave a common language to efforts that extracted intensity, texture, shape, and higher-order spatial features from routine scans and linked them to diagnosis, prognosis, treatment response, and tumor heterogeneity.
This mattered because radiomics promised to make existing clinical imaging infrastructure serve a second purpose: not only visual assessment, but large-scale data generation. If robust features could be extracted reproducibly from ordinary images, then retrospective hospital archives and prospective clinical workflows could become sources of biomarkers without requiring new invasive assays for every patient. In oncology especially, this made it newly plausible to study spatial heterogeneity across whole tumors and metastatic sites, complementing biopsy-based genomics, which samples only limited tissue regions.
The paper also anticipated the next phase of imaging AI. Later deep learning systems often learned features directly from pixels or voxels rather than relying only on handcrafted radiomic descriptors, but the conceptual bridge was similar: medical images could be integrated with clinical, molecular, and outcomes data to support prediction and decision-making. Its enduring influence comes from making that paradigm explicit while also naming the central challenges: standardization, validation, reproducibility, and clinical utility. Those constraints shaped much of the subsequent work in radiomics, radiogenomics, imaging biomarkers, and multimodal precision oncology.
Abstract¶
In the past decade, the field of medical image analysis has grown exponentially, with an increased number of pattern recognition tools and an increase in data set sizes. These advances have facilitated the development of processes for high-throughput extraction of quantitative features that result in the conversion of images into mineable data and the subsequent analysis of these data for decision support; this practice is termed radiomics. This is in contrast to the traditional practice of treating medical images as pictures intended solely for visual interpretation. Radiomic data contain first-, second-, and higher-order statistics. These data are combined with other patient data and are mined with sophisticated bioinformatics tools to develop models that may potentially improve diagnostic, prognostic, and predictive accuracy. Because radiomics analyses are intended to be conducted with standard of care images, it is conceivable that conversion of digital images to mineable data will eventually become routine practice. This report describes the process of radiomics, its challenges, and its potential power to facilitate better clinical decision making, particularly in the care of patients with cancer.
Related¶
- cite → Textural Features for Image Classification — The radiomics paper cites Haralick texture features as an early image-quantification method for extracting structured data from pixel patterns.
- cite → Intratumor Heterogeneity and Branched Evolution Revealed by Multiregion Sequencing — The radiomics paper cites multiregion sequencing evidence for intratumor heterogeneity as a biological rationale for imaging-based spatial phenotyping.
- enables ← Textural Features for Image Classification — Haralick texture features introduced quantitative image descriptors that radiomics generalized into high-throughput medical-image feature extraction.