Deep-learning metal artifact reduction achieved sub-3% dose errors in head-and-neck radiotherapy
Retrospective DL-MAR studies achieved dose errors below 3%, but no prospective clinical validation has yet been reported.
Tag
deep-learning
OmicsMap improved CT-based distant metastasis prediction across head and neck cancer cohorts
Modeling relationships between CT radiomic features improved distant-metastasis prediction over conventional radiomics across independent head and neck cancer cohorts.
Deep learning predicted distant metastasis from CT and dose in head and neck cancer
A model integrating planning CT, 3D dose and tumor contours achieved strong external prediction of distant metastasis but weaker local-recurrence performance.
Multimodal AI predicted radiation proctitis across different CT devices with 0.865 AUC
Combined radiomics, foundation-model and clinical features predicted grade ≥2 radiation proctitis with 0.865 AUC on an independent imaging device.
Deep learning generated CT-free PET with near-zero SUV bias in NSCLC
Emission-only PET synthesis produced 0.72% SUVmax bias and 99.2% contrast recovery in an independent NSCLC test cohort.
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%.
Deep-learning dose prediction generalized across six head and neck radiotherapy centres
A model trained at one centre closely predicted doses in 560 head and neck plans from six external institutions.
Interpretable autoencoder detected anomalous IMRT plans with high discrimination
An autoencoder identified anomalous IMRT plans with an area under the curve of 0.98 and explained which parameters drove each alert.
MRI deep learning identified oral tongue cancers associated with postoperative radiotherapy benefit
An MRI and pathology-based model identified a poor-risk oral tongue cancer group in which postoperative radiotherapy was associated with improved survival.
Deep learning reconstructed one-minute gated CBCT for pancreatic radiotherapy
Deep-learning reconstruction produced pancreatic gated cone-beam computed tomography comparable with conventional imaging while reducing estimated acquisition time and imaging dose.
CT imaging features improved local and distant control prediction in nasopharyngeal carcinoma
Planning CT features added prognostic information for local and distant control but did not significantly improve overall or progression-free survival prediction.
Bidirectional GAN preserved CT numbers and dose accuracy in head-and-neck planning CT
BiHU-GAN generated synthetic contrast and non-contrast CT with high contour agreement and at least 95% gamma pass rates at 2%/2 mm.
Feature-wise scanning network improved liver radiotherapy segmentation efficiency
Feature-wise scanning Mamba improved liver radiotherapy segmentation accuracy while reducing model size and inference time.
Deep learning dose prediction met clinical constraints in esophageal radiotherapy planning
A conditional generative adversarial network predicted clinically acceptable esophageal radiotherapy dose distributions, but systematic target underdosage limited clinical use.
Deep learning stopping power maps supported magnetic resonance-only proton planning for brain tumors
Synthetic stopping power maps showed small target dose differences, but residual range and low-dose gamma errors still require further validation.