Robustness of Segmentation Models#
9: The impact of U-Net architecture choices and skip connections on the robustness of segmentation across texture variations#
This work was published in Computers in Biology and Medicine (CiBM). More details about this work is included in Chapter 9.
10: Do We Really Need that Skip-Connection? Understanding Its Interplay with Task Complexity#
This work was presented as a poster at MICCAI 2023.
11: How do 3D image segmentation networks behave across the context versus foreground ratio trade-off?#
This work was presented as a poster at Medical Imaging meets NeurIPS 2022.
Ongoing connection: These architecture studies continue as a broader robust-segmentation program and inform personalized contour models whose output diversity should reflect genuine ambiguity rather than instability.
Contents
- The impact of U-Net architecture choices and skip connections on the robustness of segmentation across texture variations
- Do We Really Need that Skip-Connection? Understanding Its Interplay with Task Complexity
- How do 3D image segmentation networks behave across the context versus foreground ratio trade-off?