Mask-TS Net: Mask Temperature Scaling Uncertainty Calibration for Polyp Segmentation
Yudian Zhang, Chenhao Xu, Kaiye Xu, Haijiang Zhu
TL;DR
The paper addresses unreliable probability estimates in polyp segmentation and proposes a post-hoc calibration framework, Mask-TS, that learns per-pixel temperature scaling guided by a mask of potential lesion regions. The method uses a four-branch calibration network with Mask-Loss and Mask-TS to calibrate lesion-area probabilities without sacrificing accuracy. Empirical results on polyp datasets show improved calibration metrics (ECE, MCE, SCE, ACE) and produce uncertainty maps that align with segmentation errors, offering clinicians practical uncertainty guidance. This work advances reliable uncertainty estimation in medical image segmentation by targeting lesion regions and providing a deployable post-hoc calibration approach.
Abstract
Lots of popular calibration methods in medical images focus on classification, but there are few comparable studies on semantic segmentation. In polyp segmentation of medical images, we find most diseased area occupies only a small portion of the entire image, resulting in previous models being not well-calibrated for lesion regions but well-calibrated for background, despite their seemingly better Expected Calibration Error (ECE) scores overall. Therefore, we proposed four-branches calibration network with Mask-Loss and Mask-TS strategies to more focus on the scaling of logits within potential lesion regions, which serves to mitigate the influence of background interference. In the experiments, we compare the existing calibration methods with the proposed Mask Temperature Scaling (Mask-TS). The results indicate that the proposed calibration network outperforms other methods both qualitatively and quantitatively.