A combined automatic organ at risk contouring and synthetic Computed Tomography solution for pelvic Magnetic Resonance-only radiotherapy
Abstract
Background and Purpose: Magnetic Resonance (MR)-only radiotherapy improves treatment accuracy and efficiency but requires a synthetic Computed Tomography (sCT) for dose calculations. Automatic MR-based Organs At Risk (OARs) contouring could further improve consistency and efficiency. A combined automatic MR OAR contouring and sCT solution has been developed, with each component separately validated. This study aimed to evaluate this combined MR-only workflow against a manual MR-CT workflow for pelvic cancers. Materials and Methods: Radiotherapy MR and CT scans were acquired for 20 patients (10 prostate, 4 rectum, 6 anus cancer). The MR-only workflow used sCTs and automatic contours for the bladder, bowel bag, femoral heads, rectum, penile bulb, prostate, seminal vesicles and urethra along with manual target contours to optimise a MR-only plan. Doses were compared to the current clinical standard of deformably registered CT and manual MR OAR contours (MR-CT). Results: Mean dose differences were ≤ 1 Gy ; 1 percentage point relative volume for all OARs except the bowel bag. Mean absolute dose differences were also ≤ 1 Gy for all OARs except bowel bag in ano-rectal patients and except bowel bag, bladder V 50 Gy , V 40 Gy , V 30 Gy and rectum V 30 Gy in prostate patients. Conclusions: The automatic MR OAR contouring and sCT workflow had similar dose differences to manual inter-observer variability for all OARs except the lower dose bladder and rectum volumes (prostate patients only) and bowel bag (prostate and ano-rectal patients). This combined OAR contouring and MR-only workflow could enable more efficient and consistent pelvic radiotherapy treatment.
Fields of Study & Subjects
Advanced Radiotherapy Techniques
Prostate Cancer Diagnosis and Treatment
MRI in cancer diagnosis
Publication details
- Journal / source
- Physics and Imaging in Radiation Oncology
- Publisher
- Elsevier BV
- Volume
- 37
- Pages
- 100918–100918
- Publication date
- 1 January 2026
- Language
- EN