radiomic

Radiomics is a field within medical imaging that focuses on the high-throughput extraction and analysis of large numbers of quantitative features from medical images to characterize tumor phenotypes and tissue properties. The goal is to identify imaging biomarkers that reflect biology and can inform prognosis, treatment planning, and monitoring, beyond what is visible to the naked eye.

A typical radiomics workflow includes image acquisition, segmentation of the region of interest, extraction of quantitative features from the segmented region, feature selection to reduce redundancy, and building predictive models using statistical methods or machine learning. Features commonly fall into first-order statistics (intensity distributions), texture descriptors (such as gray-level co-occurrence and run-length features), shape metrics, and, in advanced workflows, wavelet or other transform-based features. Some radiomic approaches also incorporate features learned by deep neural networks, though these are usually distinct from classic radiomics.

Radiomics studies predominantly use computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET), alone or in combination, applied to tumors such as cancers but also in other diseases.

Applications include prognosis estimation, prediction of therapy response, lesion characterization, and, in some cases, correlating imaging features with genomic or molecular data (radiogenomics).

Challenges include variability in imaging protocols and scanners, which can affect feature stability; the need for standardization and robust validation; risks of overfitting with small datasets; and the requirement for external validation before clinical adoption. Efforts by initiatives like the Image Biomarker Standardisation Initiative (IBSI) aim to improve reproducibility and reporting.

Radiomics is an active area of research with growing potential to contribute to precision medicine, but its clinical utility depends on rigorous validation, standardization, and integration into decision-making workflows.