SUPER-Net: Trustworthy Image Segmentation via Uncertainty Propagation in Encoder-Decoder Networks
Abstract
Deep Learning (DL) holds great promise in reshaping the industry owing to its precision, efficiency, and objectivity. However, the brittleness of DL models to noisy and out-of-distribution inputs is ailing their deployment in sensitive fields. Current models often lack uncertainty quantification, providing only point estimates. We propose SUPER-Net, a Bayesian framework for trustworthy image segmentation via uncertainty propagation. Using Taylor series approximations, SUPER-Net propagates the mean and covariance of the model's posterior distribution across nonlinear layers. It generates two outputs simultaneously: the segmented image and a pixel-wise uncertainty map, eliminating the need for expensive Monte Carlo sampling. SUPER-Net's performance is extensively evaluated on MRI and CT scans under various noisy and adversarial conditions. Results show that SUPER-Net outperforms state-of-the-art models in robustness and accuracy. The uncertainty map identifies low-confidence areas affected by noise or attacks, allowing the model to self-assess segmentation reliability, particularly when errors arise from noise or adversarial examples.
Cite
@article{arxiv.2111.05978,
title = {SUPER-Net: Trustworthy Image Segmentation via Uncertainty Propagation in Encoder-Decoder Networks},
author = {Giuseppina Carannante and Nidhal C. Bouaynaya and Dimah Dera and Hassan M. Fathallah-Shaykh and Ghulam Rasool},
journal= {arXiv preprint arXiv:2111.05978},
year = {2025}
}
Comments
Accepted in Pattern Recognition. This is the author's accepted manuscript. Licensed under CC BY-NC-ND. The Version of Record is available at https://doi.org/10.1016/j.patcog.2025.112503