OCTAVIUS 4D achieved 98.3% gamma passing for MR-LINAC SBRT QA
Across 40 Unity SBRT plans, OCTAVIUS 4D achieved 98.3% mean local gamma passing with sub-degree angular agreement.
Tag
technical-relevance
Deep learning reduced combined truncation and metal artifacts in half-detector CBCT
HD-TMAR improved metal-region PSNR from 25.5 to 35.5 dB in simulated half-detector CBCT while preserving dental anatomy.
MR-only CyberKnife planning achieved near-equivalent dosimetry for brain metastases
Synthetic CT reproduced CyberKnife target dosimetry closely and enabled submillimetre translational image guidance, although rotational errors were more variable.
Task-specific deep learning reduced CT metal artifacts in cervical brachytherapy
Task-specific deep learning halved pelvic artifact burden, improved applicator reconstruction to 0.1 mm, and shortened organ contouring by up to 40%.
Automated proton planning reproduced clinical head and neck NTCP estimates within minutes
Automated proton plans reproduced clinical dysphagia and xerostomia estimates closely while reducing optimization time from days to about one hour.
Prostate target expansion identified fractions most likely to benefit from online re-optimization
Without online re-optimization, 65.3% of prostate fractions missed 95% target coverage, with later target expansion strongly linked to coverage loss.
Individual target optimization reduced brain dose in single-isocenter multi-metastasis radiosurgery
Individual target optimization lowered normal-brain V12 and improved dose gradients while maintaining over 99% coverage, at the cost of higher monitor units.
On Ethos, intensity-modulated radiotherapy generated faster and spared several pelvic organs better than arc therapy
Nine-field intensity-modulated radiotherapy shortened overall planning-plus-delivery time and improved several pelvic dose metrics compared with volumetric modulated arc therapy.
Deep learning generated deliverable lymphoma intensity-modulated plans but remained dosimetrically inferior
Deep learning produced machine-deliverable lymphoma plans across heterogeneous anatomy, but target coverage and hot spots remained inferior to reference planning.
Audit logs measured artificial intelligence contour correction time with 0.93 concordance
Treatment-planning audit logs closely reproduced manual contour-edit timing and exposed major structure-specific differences in correction workload.
Diaphragm motion was less reliable than fiducials for markerless tumor tracking
Tumor–diaphragm geometry varied by approximately 4.5 mm during breathing, substantially more than tumor–fiducial relationships.
Eye tracking and imaging features predicted uncertainty in manual brain tumor contours
Behavioral and image-derived features explained 39% of interobserver contour uncertainty during manual brain tumor delineation.
Heart dose rises during esophageal IMPT but adaptive replanning limits the increase
Mean heart dose increased during esophageal IMPT, while heart-dose–triggered replanning reduced the excess dose in patients requiring adaptation.
Real-world brain MR-only implementation exposed frequent synthetic CT artifacts
Synthetic CT artifacts were common during brain MR-only implementation, highlighting workflow QA and patient-selection issues not captured by controlled validation.
Commercial AI synthetic CT showed good soft-tissue agreement but important high-density failures
AI synthetic CT preserved soft-tissue HU and geometry well, while bone, air interfaces and implanted devices remained important failure modes.