Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Adaptation of Weakly Supervised Localization in Histopathology by Debiasing Predictions

Alexis Guichemerre, Banafsheh Karimian, Soufiane Belharbi, Natacha Gillet, Nicolas Thome, Pourya Shamsolmoali, Mohammadhadi Shateri, Luke McCaffrey, Eric Granger

Abstract

Weakly Supervised Object Localization (WSOL) models enable joint classification and region-of-interest localization in histology images using only image-class supervision. When deployed in a target domain, distributions shift remains a major cause of performance degradation, especially when applied on new organs or institutions with different staining protocols and scanner characteristics. Under stronger cross-domain shifts, WSOL predictions can become biased toward dominant classes, producing highly skewed pseudo-label distributions in the target domain. Source-Free (Unsupervised) Domain Adaptation (SFDA) methods are commonly employed to address domain shift. However, because they rely on self-training, the initial bias is reinforced over training iterations, degrading both classification and localization tasks. We identify this amplification of prediction bias as a primary obstacle to the SFDA of WSOL models in histopathology. This paper introduces \sfdadep, a method inspired by machine unlearning that formulates SFDA as an iterative process of identifying and correcting prediction bias. It periodically identifies target images from over-predicted classes and selectively reduces the predictive confidence for uncertain (high entropy) images, while preserving confident predictions. This process reduces the drift of decision boundaries and bias toward dominant classes. A jointly optimized pixel-level classifier further restores discriminative localization features under distribution shift. Extensive experiments on cross-organ and -center histopathology benchmarks (glas, CAMELYON-16, CAMELYON-17) with several WSOL models show that SFDA-DeP consistently improves classification and localization over state-of-the-art SFDA baselines. {\small Code: \href{https://anonymous.4open.science/r/SFDA-DeP-1797/}{anonymous.4open.science/r/SFDA-DeP-1797/}}

Adaptation of Weakly Supervised Localization in Histopathology by Debiasing Predictions

Abstract

Weakly Supervised Object Localization (WSOL) models enable joint classification and region-of-interest localization in histology images using only image-class supervision. When deployed in a target domain, distributions shift remains a major cause of performance degradation, especially when applied on new organs or institutions with different staining protocols and scanner characteristics. Under stronger cross-domain shifts, WSOL predictions can become biased toward dominant classes, producing highly skewed pseudo-label distributions in the target domain. Source-Free (Unsupervised) Domain Adaptation (SFDA) methods are commonly employed to address domain shift. However, because they rely on self-training, the initial bias is reinforced over training iterations, degrading both classification and localization tasks. We identify this amplification of prediction bias as a primary obstacle to the SFDA of WSOL models in histopathology. This paper introduces \sfdadep, a method inspired by machine unlearning that formulates SFDA as an iterative process of identifying and correcting prediction bias. It periodically identifies target images from over-predicted classes and selectively reduces the predictive confidence for uncertain (high entropy) images, while preserving confident predictions. This process reduces the drift of decision boundaries and bias toward dominant classes. A jointly optimized pixel-level classifier further restores discriminative localization features under distribution shift. Extensive experiments on cross-organ and -center histopathology benchmarks (glas, CAMELYON-16, CAMELYON-17) with several WSOL models show that SFDA-DeP consistently improves classification and localization over state-of-the-art SFDA baselines. {\small Code: \href{https://anonymous.4open.science/r/SFDA-DeP-1797/}{anonymous.4open.science/r/SFDA-DeP-1797/}}
Paper Structure (6 sections, 5 equations, 4 figures, 2 tables)

This paper contains 6 sections, 5 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Proposed SFDA-DeP Method
  4. Results and Discussion
  5. Experimental Methodology:
  6. Conclusion

Figures (4)

  • Figure 1: (a) Source-only inference shows a strong predictive bias toward the cancer class when transferring from GlaS (source) to CAMELYON16 and CAMELYON17 (target). (b) Using CAMELYON16 as source yields more balanced predictions across targets (CAMELYON17 and GlaS). (c) SFDA-DeP is shown to reduce predictive bias across centers. (d) In contrast, traditional SFDA, such as SFDA-DE amplifies bias, often collapsing to a dominant class. (The dashed line indicates the ideal 50% predictive balance for these datasets.)
  • Figure 2: Effect of dynamic resampling frequency during adaptation with PixelCAM.
  • Figure 3: Impact on performance of (a) the localization loss during adaptation, and of (b) dynamic resampling versus static. (GlaS → CAMELYON17).
  • Figure 4: Activation maps produced by the PixelCAM WSOL model after SFDA-EeP and state-or-the-art SFDA methods (GlaS$\rightarrow$CAMELYON17-3).