Randomized trial finds no efficiency gain from AI-assisted head and neck tumor contouring
In 83 patients, AI-assisted tumor contouring did not shorten clinical review or contouring time, with a possible increase in nodal contour editing.
In 83 patients, AI-assisted tumor contouring did not shorten clinical review or contouring time, with a possible increase in nodal contour editing.
A self-supervised model detected clinical pelvic MRI anomalies with AUC 0.95, but missed all five hydrogel spacers and lacked downstream clinical validation.
A critical review of 11 studies found major workflow gains from AI-assisted contouring and planning, but limited evidence of sustainable implementation in resource-constrained settings.
Across 24 expert sessions, AI-assisted scribble editing reduced median contour refinement time by 70% while maintaining similar final contour quality.
Among 702 prostate RT patients, an optimized language model reproduced manual NCCN risk groups in 96.2%, with most errors between intermediate-risk categories.
AI-generated partial breast plans met all constraints in six of seven end-to-end cases, but fixed beam geometry caused frequent heart-dose violations.
Multimodal LLM prediction reached an AUC of 0.79, but AI-generated MRI reports showed clinically important discordances requiring expert review.
Three commercial AI models produced broadly usable head-and-neck contours, reducing review time to five minutes, but anatomical definitions differed importantly between vendors.
Ten contemporary multimodal deep-learning studies showed inconsistent external performance, calibration and reporting, limiting clinical use of personalized toxicity prediction.
External validation achieved AUC 0.986, while modeled risk-adapted surveillance reduced visits by 90.0–98.5% and identified every observed failure.
AI can automate particle planning, dose prediction, adaptation and quality assurance, but limited data and validation still constrain clinical adoption.
A model integrating planning CT, 3D dose and tumor contours achieved strong external prediction of distant metastasis but weaker local-recurrence performance.
ASTuTE biomarker testing changed 27.5% of shared decisions and reduced planned short-term ADT use from 37% to 12.5%.
A scoping review maps generative AI across imaging, documentation, QA and decision support while emphasizing major safety and governance gaps.
AI synthetic CT preserved soft-tissue HU and geometry well, while bone, air interfaces and implanted devices remained important failure modes.