Deliverability-aware deep learning substantially improves automated IMRT planning for lymphoma
Adding fluence supervision and cycle consistency substantially improved machine-deliverable lymphoma IMRT plans compared with dose-only deep learning.
Adding fluence supervision and cycle consistency substantially improved machine-deliverable lymphoma IMRT plans compared with dose-only deep learning.
Permanent palladium-103 breast seed implantation achieved 100% 10-year local control with no grade ≥3 toxicity in selected early breast cancer.
Adaptive IMPT improved cumulative target coverage and reduced heart dose without increasing lung dose in a 192-patient retrospective cohort.
Serial MR-linac imaging showed frequent glioblastoma target changes, with some extending beyond the planned CTV and 95% isodose.
After head and neck RT, 43% of patients lost teeth and higher oral cavity dose was associated with tooth loss.
Five randomized trials from ASTRO 2026, covering SCLC, prostate cancer, liver tumors and brain metastases—with key findings and limitations.
Across 212 CBCTs, 36.8% showed target extensions >10 mm beyond the planning PTV despite progressive overall CTV shrinkage.
An automated cervical brachytherapy QA system detected preset planning errors and achieved 98.7% mean gamma agreement with Oncentra.
ZeeFree improved 4DCT image quality and anatomical correctness during irregular breathing without degrading already artifact-free scans.
Long-course chemoradiation caused greater preoperative symptom burden, but patient-reported bowel, urinary and overall outcomes converged after surgery.
T1 MRI with R-MRmr reached AUC 0.96 for early xerostomia, while late sticky-saliva prediction remained substantially weaker.
Across 18 postoperative brain metastasis cavities, GammaTile generated consistent peak–valley heterogeneity, with the nominal 60 Gy prescription approximating D90.
A 33 Gy/6-fraction boost regimen improved plexus-related symptoms in 84% of evaluable patients and achieved 74% freedom from local progression at one year.
AI can automate major components of CBCT-guided adaptation, but validated triggers, end-to-end QA and patient-outcome evidence remain limited.
Lower dose to diffusion-defined contralateral parotid subvolumes showed early associations with xerostomia-related outcomes, but most dose–toxicity relationships were not significant.