Beyond scalar losses: calibrating segmentation models via gradient vector field surgery
For medical imaging applications requiring calibrated predictions (e.g., tumor resection margins), this provides a simple fix to a known bottleneck in region-based losses.
This work addresses miscalibration in segmentation models trained with region-based losses, proposing a gradient vector field surgery that scales gradient magnitude with prediction error. The method reduces overconfidence while maintaining accuracy across 2D and 3D medical segmentation tasks.
Region-based loss functions, such as the Dice loss, have established themselves as the de facto standard for highly class- and region-imbalanced segmentation tasks. However, models trained using region-based loss functions are notoriously miscalibrated and typically yield over-confident predictions. In medical imaging applications, such as defining tumor resection margins, this miscalibration is hindering clinical adoption. In this work, we outline a novel gradient perspective on this overconfidence and show how it affects region-based loss functions. We propose a "surgery" on the gradient vector field as a simple, yet effective intervention to mitigate calibration issues. This surgery adds a factor to the loss's partial derivative, scaling the gradient's magnitude linearly with the prediction error. In empirical evaluations across 2D and 3D medical segmentation tasks, we demonstrate the effectiveness of this intervention while maintaining high prediction accuracy when used in conjunction with any region-based loss function.