Abstract Radiomics represents a significant step forward in oncology, as it converts standard medical imaging into high-dimensional, quantitative data. These data include tumour phenotype and biological diversity that extends beyond what can be visually observed. Recent improvements in artificial intelligence, machine learning, and standardising imagery have accelerated its evolution from a tool for exploratory study. It is increasingly beneficial in precision oncology clinical environments. This study reviews advancements in radiomics and radiogenomics concerning methodological workflows, diagnostic, prognostic, and predictive applications, as well as clinical translation. Critical analysis includes image acquisition, harmonisation, segmentation, model development, validation, feature extraction, and interpretability. Radiomics is increasingly significant in the research of lung, brain, gynaecological, hepatobiliary, and gastrointestinal cancers. Radiomics has the potential to forecast outcomes, assess therapy responses, and customise therapies. Radiogenomics detects molecular alterations pertinent to targeted therapy and immunotherapy without invasive procedures. Radiomics can offer dynamic, biology-based treatment strategies due to advancements in tailored radiotherapy, toxicity forecasting, and delta-radiomics. Reproducibility, algorithmic transparency, ethical governance, data heterogeneity, and regulatory oversight are also analysed. The challenges are examined in accordance with contemporary international criteria. Radiomics presents considerable potential as a clinically pertinent decision-support instrument in precision oncology. Ensuring methodological rigour, external validation, explainable AI, and ethical integration is crucial for the dependable translation of findings from research environments. The translation should take place before its implementation in routine clinical practice.
Thenuan et al. (Thu,) studied this question.