Meeting Banner
Abstract #2832

Deep learning for prostate zonal segmentation robust to multicenter data

Simona Turco1, Hubert Blach1, Catarina Dinis Fernandes1, Jelle Barentsz2, Stijn Heijmink 3, Hessel Wijkstra1, and Massimo Mischi1
1Electrical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands, 2Radiology, Radboud university medical center, Nijmegen, Netherlands, 3Radiology, Netherlands Cancer Institute, Amsterdam, Netherlands


Prostate zonal segmentation is an important step for automated PCa diagnosis, MRI-guidedradiotherapy and focal treatment planning. Here we proposed a multi-channel U-Net for automatic prostate zonal segmentation, able to include multiple MRI sequences. Using a small, multicenter, multiparametric MRI dataset, we investigated its robustness towards the acquisition protocol and whether additional imaging sequences improve segmentation performance. Our results show that T2-weighted imaging alone is sufficient for successful prostate zonal segmentation. Despite using a small multicenter dataset, the models were robust towards the acquisition protocol and the performance was comparable to that obtained with larger datasets from a single institute.

This abstract and the presentation materials are available to members only; a login is required.

Join Here