JeeSuk Chang: A multi-center trial – great efforts from the KROG team
JeeSuk Chang,
“1. Check out our new paper exploring the potential clinical utility of deep learning auto-contouring for clinical target volume, organs at risk in breast cancer Radiation Therapy Quality Assurance within a multi-center trial. Great efforts from the KROG team and credit to the first author, Chloe Min Seo Choi. 31 centers participated.
For the article click here.
2. Over 800 pairwise comparisons showed significant improvement in inter-observer variation with auto-contour intervention. Participants’ contours aligned with auto-contour reference shapes in 2D and 3D, pinpointing common adjustments in local anatomical regions.
3. Cluster analysis based on CTV volumes classified contouring styles, potentially laying the groundwork for reducing inter-observer variations. For instance, certain centers (red) inclined to contour larger than average were identified.
4. Along with education, audits, site credentialing, and peer reviews, using auto-contours for CTV and OARs can be a novel approach to minimize RT protocol deviations in multi-center trials.”
Source: JeeSuk Chang/Twitter