S2795
RTT - RTT contouring, target definition, and treatment planning
ESTRO 2026
Group, Northwood, United Kingdom. 3 Department of Clinical Oncology, The Christie NHS Foundation Trust, Manchester, United Kingdom. 4 Department of Oncology, The Shrewsbury and Telford Hospital NHS Trust, Shrewsbury, United Kingdom. 5 Department of Oncology, Oxford University Hospitals NHS Foundation Trust, Oxford, United Kingdom. 6 Department of Oncology, University of Oxford, Oxford, United Kingdom. 7 Leeds Institute of Medical Research, University of Leeds, Leeds, United Kingdom. 8 Department of Oncology, Rigshospitalet & Technical University Hospital of Greater Copenhagen, Copenhagen, Denmark. 9 Department of Oncology, Leeds Teaching Hospitals NHS Trust, Leeds, United Kingdom
CUBE MRI images (CUBE), supporting MR-based radiotherapy (RT) planning. The goal of this work was to assess the performance and limitations of this method on 2D T2 FLEX images (FLEX). Material/Methods: Forty-seven FLEX scans (8 female, 39 male, Umeå University Hospital, Sweden) were processed, and compared with the results of 55 CUBE scans (10 female, 45 male, from US and UK sources). Six unisex (pelvic-body, bowel-bag, bladder, rectum, and bilateral femoral-heads) and 4 male (prostate, seminal-vesicles, urethra, penile-bulb) OARs were segmented using MRCDL. Contours were rated on a 5-point Likert scale: 5–3 indicated clinically usable (no, minor, or some correction), while 2–1 indicated unusable (1 requiring full recontouring). Results: On FLEX images, the average Likert score was 3.6. Organ-specific scores were pelvic-body 4.0, bowel-bag 3.5, bladder 3.8, prostate 3.1, seminal-vesicles 2.8, urethra 3.5, penile-bulb 3.4, rectum 3.4, and femoral- heads 4.3. Of the 424 contours, 88% (373) were clinically acceptable: 16% (68) required no correction, 46% (197) minor, and 25% (108) some correction. The results of the 2 datasets were compared by subtracting average CUBE scores from FLEX scores. The differences were: overall -0.1, pelvic-body -0.4, bowel-bag 0.0, bladder -0.1, prostate -0.3, seminal- vesicles -0.5, urethra 0.0, penile-bulb 0.1, rectum 0.0, femoral-heads 0.1. Significant differences occurred for seminal-vesicles (p=0.017), likely due to higher noise and contrast variability in FLEX, and pelvic-body (p=0.011), possibly from edge-of-field blurring, motion artifacts, and inclusion of the patient’s arm in 7 cases.
Digital Poster 4047 Multi-sequence Evaluation of MR Pelvic Organ Segmentation Zsófia Karancsi 1,2 , Bernadett Kolozsvári 1 , Mikael Bylund 1 , Eszter K. Kiss 1 , Eszter Szabo 1 , Borbála Deák- Karancsi 1 , László Ruskó 1 1 Science and Technology, GE HealthCare, Budapest, Hungary. 2 Department of Pathology Forensic and Insurance Medicine, Semmelweis University, Budapest, Hungary Purpose/Objective: MR Contour DL (MRCDL) is an FDA-approved method for segmenting pelvic organs at risks (OARs) on 3D T2
Subject-specific analysis showed significantly reduced accuracy for benign prostate hyperplasia (n=5, prostate: 2.2; seminal vesicles: 2.0), while inclusion of the patient’s arm lowered pelvic body scores (3.3). Other conditions - such as empty bladder, bladder diverticula, dilated urethra, and stoma - did not adversely affect the segmentation. Intra-observer agreement across CUBE and FLEX scores was substantial, with a weighted Cohen’s Kappa of 0.727
Made with FlippingBook - Share PDF online