Title : The role of Artificial Intelligence in enhancing radiotherapy practice: A scoping review
Abstract:
Background: Artificial intelligence (AI) is emerging, with significant potential to enhance radiotherapy practice in practically every step of the way. Radiotherapy involves complex labor intensive and operator dependent preparation before treatment. This scoping review updates the possible AI usage, focusing on medical decision support to choose radiotherapy (Bayesian networks comparing radiotherapy with other treatment modalities), synthetic computed tomography, image registration/fusion, target/organ at risk delineation, treatment planning, evaluation, quality assurance (QA), adaptive treatments, workflow automation and patient engagement tools.
Methods: A search of four databases (Scopus, ScienceDirect, PubMed, and CINAHL) was conducted for original peer reviewed studies (2016/1/1-2026/1/31), using PRISMA-ScR (Preferred Reporting Items for Systematic reviews and Meta Analyses extension for Scoping Reviews). Nineteen studies met the inclusion criteria.
Results: Head and neck site is the most commonly studied (32%). AI assisted tools can support clinical decision making (choice of proton versus photons or stereotactic radiosurgery versus fractionated radiotherapy by predicting toxicities), reduce administrative burden (analysis of referral pattern, managing wait times, tracking patient on treatment reviews and post-treatment follow-ups), and improve care quality (analyses of motion, set-up errors, toxicities, failure pattern). One study used questionnaires demonstrating strong patient support for AI use. Another study was on two types of uncertainty, aleatoric (noise inherent in the data) and epistemic (incomplete information). Significant gaps remain in evaluating uncertainties (only 23% of studies explicitly distinguishing between aleatoric and epistemic uncertainty) and practical implementation. NRG (National Surgical Adjuvant Breast and Bowel Project/Radiation Therapy Oncology Group/Gynecologic Oncology Group) will evaluate clinical utilization and potential of commercial AI auto segmentation tools, challenges and limitations for clinical trials in the future.
Conclusion: This review summarizes the use of AI in radiotherapy decisions, predictive analytics and workflow. It highlights the need for more rigorous, large scale studies and provides recommendations to guide future research, policy, and practice in integrating AI into radiotherapy.
