Multicentre evaluation of deep learning CT autosegmentation of the head and neck region for radiotherapy.
Authors
Affiliations (16)
Affiliations (16)
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore, 158683, Singapore.
- Oncology Academic Clinical Programme, Duke-NUS Graduate Medical School, Singapore, 169857, Singapore.
- Division of Radiation Oncology, National Cancer Centre Singapore, Singapore, 158683, Singapore. [email protected].
- Oncology Academic Clinical Programme, Duke-NUS Graduate Medical School, Singapore, 169857, Singapore. [email protected].
- Division of Physics and Applied Physics, School of Physical and Mathematical Science, Nanyang Technological University, Singapore, Singapore. [email protected].
- MVision Ai, c/o Terkko Health hub, FI-00290, Helsinki, Finland.
- Docrates Cancer Center, 00180, Helsinki, Finland.
- Department of Radiotherapy, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Rotterdam, Netherlands.
- Kuopio University Hospital, Center of Oncology, FI-70210, Kuopio, Finland.
- North Estonia Medical Centre, 13419, Tallinn, Estonia.
- Oulu University Hospital, Department of Oncology and Radiotherapy, 90220, Oulu, Finland.
- Research Unit of Health Sciences and Technology, Faculty of Medicine, University of Oulu, 90220, Oulu, Finland.
- Turku University Hospital, Department of Oncology and Radiotherapy, FI-20521, Turku, Finland.
- Turku University Hospital, Department of Medical Physics, FI-20521, Turku, Finland.
- University of Helsinki, Faculty of Science, Department of Physics University of Helsinki, FI-00014, Helsinki, Finland.
- Division of Medical Sciences, National Cancer Centre Singapore, Singapore, 169610, Singapore.
Abstract
This is a multi-institutional study to evaluate a head-and-neck CT auto-segmentation software across seven institutions globally. 11 lymph node levels and 7 organs-at-risk contours were evaluated in a two-phase study design. Time savings were measured in both phases, and the inter-observer variability across the seven institutions was quantified in phase two. Overall time savings were found to be 42% in phase one and 49% in phase two. Lymph node levels IA, IB, III, IVA, and IVB showed no significant time savings, with some centers reporting longer editing times than manual delineation. All the edited ROIs showed reduced inter-observer variability compared to manual segmentation. Our study shows that auto-segmentation plays a crucial role in harmonizing contouring practices globally. However, the clinical benefits of auto-segmentation software vary significantly across ROIs and between clinics. To maximize its potential, institution-specific commissioning is required to optimize the clinical benefits.