EXPOSE: Explainable and Domain-Robust Embeddings from Pathology Vision Foundation Models using Sparse Autoencoders

Anja Witte; Maximilian Lennartz; Jan Baumbach; Guido Sauter; Stefan Bonn; Patrick Fuhlert; Marina Zimmermann

cs.CVAug 28, 2026

EXPOSE: Explainable and Domain-Robust Embeddings from Pathology Vision Foundation Models using Sparse Autoencoders

Authors: Anja Witte, Maximilian Lennartz, Jan Baumbach, Guido Sauter, Stefan Bonn, Patrick Fuhlert, Marina Zimmermann

Organizations: Institute of Medical Systems Bioinformatics, Center for Biomedical AI (bAIome), Center for Molecular Neurobiology Hamburg (ZMNH), Hamburg Center for Translational Immunology (HCTI), University Medical Center Hamburg-Eppendorf, Hamburg, Germany · Institute of Pathology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany · Institute of Computational Systems Biology, University of Hamburg, Hamburg, Germany · German Center for Child and Adolescent Health (DZKJ), Hamburg, Germany · III. Department of Medicine, University Medical Center Hamburg-Eppendorf, Hamburg, Germany

Abstract

Vision Foundation Models (VFMs) are widely used in computational pathology but remain sensitive to domain shifts arising from variations in staining, tissue preparation, and scanner hardware. A key limitation is that VFM embeddings entangle biological with domain-specific information, hindering cross-domain generalization. We propose Explainable Probing of Cross-Domain Sparse Embeddings (EXPOSE), a framework that uses Sparse Autoencoders (SAEs) as an explainable bottleneck to identify and suppress domain-specific components in VFM embeddings. We train a sparse representation of VFM features, use a linear classifier to identify domain-specific latent dimensions, and mask these features prior to downstream relapse prediction without retraining the backbone model. Experiments on a large prostate cancer dataset with multiple acquisition domains show that SAE features capture both domain- and task-specific information, which are partially disentangled in the latent space. Removing domain-specific features improves cross-domain performance and increases embedding robustness as measured by the Domain Robustness Index (DoRI). Code is available at https://github.com/imsb-uke/expose .

Figures & tables

Figure 1: Overview of expose . (A) An sae is trained in an unsupervised manner on vfm patch embeddings of two domains to learn cross-domain and disentangled representations. (B) The sae encoded features are mean-pooled and a linear classifier is trained to predict the domain of these slide-level representations. Feature dimensions are ranked according to their weights in the domain classifier. The features with the top- $k$ highest weights are defined as domain-specific features. (C) The mean-pooled features identified as domain-specific are set to zero. A linear classifier is trained to predict the binary downstream task of relapse prediction. The most relevant features for both domain and relapse classes are linked to visual features.

Figure 2: Evaluation of relapse prediction while masking the top-

k

domain-specific features chosen with different selection strategies based on the linear weights for domain classification. Top negative and top positive weights can be defined as relevant for first and scanner classification, respectively. Features that are relevant for both domains in the domain classifier are chosen by selecting the features with the top absolute domain weights. Validation results are shown for the evaluation of first (A) and scanner (B) dataset.

Figure 3: Visualization of highly activating images and patches for domain- and relapse-specific sparse features, respectively. For each ID, examples are chosen from the first (left panel) and scanner (right panel) subdataset separately. For the top negative (A) and top positive (B) domain-specific features, we show the three most strongly activating tma spots together with activation heatmaps. The top negative (C) and top positive (D) relapse-specific features are visualized with three highly activating patches and their surrounding tissue context.

Figure 4: Analysis of the sae embedding space. In (A) and (B) the sae feature dimension are visualized as activation delta between first and scanner . The UMAPs are coloured by the feature weights for domain classification (A), and relapse classification (B). In (C) and (D), we evaluate the Domain Robustness Index (DoRI) for different numbers of masked features. Gray and purple bars represent masked features that are chosen randomly and according to its domain weight, respectively. The dori is calculated within the embedding space of in-domain (C) and all subdatasets combined (D).

Dataset	Base	SAE	SAE ${}^{+}_{256}$	SAE ${}^{+}_{3500}$
first	68.20 $\pm$ 0.26	69.20 $\pm$ 0.41	69.39 $\pm$ 0.35	69.23 $\pm$ 0.08
scanner	68.58 $\pm$ 0.31	69.44 $\pm$ 0.39	69.75 $\pm$ 0.17	69.13 $\pm$ 0.16
spot	65.38 $\pm$ 0.56	65.53 $\pm$ 0.76	65.83 $\pm$ 0.71	67.74 $\pm$ 0.11
thin	60.48 $\pm$ 1.47	57.78 $\pm$ 0.48	57.80 $\pm$ 0.47	58.63 $\pm$ 0.15
thick	58.17 $\pm$ 0.74	58.40 $\pm$ 1.48	58.44 $\pm$ 0.69	59.62 $\pm$ 0.17
long	61.67 $\pm$ 1.09	61.79 $\pm$ 1.09	62.04 $\pm$ 0.58	62.67 $\pm$ 0.06

Table 1: Cross-domain evaluation of relapse prediction. Results show auroc5 (mean ± std) for Base, sae , and sae with masking of domain-specific sparse features (

k=256

k=3500

). Highlighted values indicate per-dataset best and second-best performance.

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Dataset	Images	Tissue Core	Slicing Thickness	Staining Time in min (H/E)	Scanner	Mag.
first	8141	1	2.5 $\mu$ m	4:00 / 1:20	Aperio	40x
scanner	8141	1	2.5 $\mu$ m	4:00 / 1:20	3DHistech	80x
spot	8181	2	2.5 $\mu$ m	4:00 / 1:20	Aperio	40x
thin	1951	1	1 $\mu$ m	4:00 / 1:20	Aperio	40x
thick	1950	1	10 $\mu$ m	4:00 / 1:20	Aperio	40x
long	1951	1	2.5 $\mu$ m	40:00 / 10:00	Aperio	40x

Appendix

Table 2: Overview of additional dataset variants derived from the first dataset. first follows the standard acquisition protocol, while other datasets introduce controlled variations in scanner, staining time, slicing thickness, or sampling location. Differences from the standard protocol are shown in bold. Tissue cores are obtained from two different regions of the prostate. first and scanner contain the registered images of the same tma spots scanned with different scanners. All other subdatasets of tissue core 1 are different slices of the same core.

Explore similar work

Robustifying pathology foundation models via fine-tuning
Jul 24, 2026Alexandre Filiot, Oskar Thaeter, Benoit Schmauch +1Computational Pathology Fine-Tuning
1Waiv · Institute of Pathology, Technical University of Munich · School of Computation, Information and Technology, Technical University of Munich
Discrepancy Minimization Improves Cross-Hospital Robustness in Digital Pathology
May 24, 2026Ben Vardi, Dana Schonberger, Yuval Friedmann +4Computational Pathology Unsupervised Domain Adaptation
Reichman University, Herzliya, Israel · Institute of Pathology, Sheba Medical Center, Ramat-Gan, Israel · Technion - Israel Institute of Technology, Haifa, Israel
Benchmarking Pathology Foundation Models for Spatial Domain Understanding
May 25, 2026Bokai Zhao, Yiyang Zhang, Yuanchi Zhu +6Pathology Foundation Models Medical Image Benchmarks
School of Artificial Intelligence, University of Chinese Academy of Sciences. · Brainnetome Center, Institute of Automation, Chinese Academy of Sciences. · Beijing Key Laboratory of Brainnetome and Brain-Computer Interface, Institute of Automation, Chinese Academy of Sciences. +2

Jul 24, 2026J/K move · Enter open · S save

Robustifying pathology foundation models via fine-tuning

Alexandre Filiot, Oskar Thaeter, Benoit Schmauch, Lionel Guillou

1Waiv · Institute of Pathology, Technical University of Munich · School of Computation, Information and Technology, Technical University of Munich

Computational Pathology Fine-Tuning Pathology Foundation Models

Open paper AskPDF arXiv

Pathology foundation models (FMs) produce powerful tile-level representations which remain sensitive to scanner and staining variability, undermining deployment across laboratories. We develop a novel fine-tuning recipe that improves the robustness of pathology FMs to acquisition factors. Applied to ten different FMs, our fine-tuning strategy consistently improves robustness for every model as well as downstream performance, with no observed trade-off. On average, it raises the PathoROB robustness index by 23% (from 0.72 to 0.87) and increases the overall cross-benchmark performance by 43% on Patho-Bench, HEST and THUNDER combined, with individual gains reaching up to 72% in robustness (Phikon-v2) and 76% in performance (Midnight-12k). We publicly release the fine-tuned versions of Phikon-v2 (Phaet) and Midnight-12k (Mascaret) at https://huggingface.co/wearewaiv/models.

Abstract

Explore similar work

Robustifying pathology foundation models via fine-tuning

Discrepancy Minimization Improves Cross-Hospital Robustness in Digital Pathology

Benchmarking Pathology Foundation Models for Spatial Domain Understanding