Domain Adaptation Without the Compute Burden for Efficient Whole Slide Image Analysis

Umar Marikkar; Muhammad Awais; Sara Atito

cs.CVMar 16, 2026

Domain Adaptation Without the Compute Burden for Efficient Whole Slide Image Analysis

Authors: Umar Marikkar, Muhammad Awais, Sara Atito

Organizations: Institute for People-Centered AI, University of Surrey · Centre of Vision, Speech and Signal Processing, University of Surrey

Abstract

Computational methods on analyzing Whole Slide Images (WSIs) enable early diagnosis and treatments by supporting pathologists in detection and classification of tumors. However, the extremely high resolution of WSIs makes end-to-end training impractical compared to typical image analysis tasks. To address this, most approaches use pre-trained feature extractors to obtain fixed representations of whole slides, which are then combined with Multiple Instance Learning (MIL) for downstream tasks. These feature extractors are typically pre-trained on natural image datasets such as ImageNet, which fail to capture domain-specific characteristics. Although domain-specific pre-training on histopathology data yields more relevant feature representations, it remains computationally expensive and fail to capture task-specific characteristics within the domain. To address the computational cost and lack of task-specificity in domain-specific pre-training, we propose EfficientWSI (eWSI), a careful integration of Parameter-Efficient-Fine-Tuning (PEFT) and Multiple Instance Learning (MIL) that enables end-to-end training on WSI tasks. We evaluate eWSI on seven WSI-level tasks over Camelyon16, TCGA and BRACS datasets. Our results show that eWSI when applied with ImageNet feature extractors yields strong classification performance, matching or outperforming MILs with in-domain feature extractors, alleviating the need for extensive in-domain pre-training. Furthermore, when eWSI is applied with in-domain feature extractors, it further improves classification performance in most cases, demonstrating its ability to capture task-specific information where beneficial. Our findings suggest that eWSI provides a task-targeted, computationally efficient path for WSI tasks, offering a promising direction for task-specific learning in computational pathology.

Figures & tables

Figure 1: Moving from ImageNet to domain specific encoders, an easy path through eWSI (radial axis is %AUC).

Encoder state	Method	Encoder initialization
Encoder state	Method	iNet-sup	iNet-SSL	Hist-SSL
Frozen:*	ABMIL	79.0 $\pm$ 4.9	-	94.5 $\pm$ 2.7
	CLAM-SB	76.3 $\pm$ 4.9	-	96.9 $\pm$ 2.4
	TransMIL	70.6 $\pm$ 7.6	-	94.3 $\pm$ 0.9
	DSMIL	77.3 $\pm$ 3.4	-	91.7 $\pm$ 0.0
	IBMIL	79.9 $\pm$ 5.0	-	95.4 $\pm$ 2.2

Table 1: Performance Comparison (% AUC) on Camelyon16 test split. (*: values reported in literature,

\star

: our standardized implementation.

\pm

0.0 : std not reported.)

	Method	TCGA					BRACS
	Method	BRCA	NSCLC	mHRD	tHRD	TNBC	BRACS
Inet-sup	ABMIL	83.7 $\pm$ 6.4	92.8 $\pm$ 2.1	67.0 $\pm$ 6.2	72.7 $\pm$ 4.2	85.7 $\pm$ 4.3	76.4 $\pm$ 2.5
	TransMIL	86.4 $\pm$ 4.9	90.3 $\pm$ 3.3	61.1 $\pm$ 3.9	72.9 $\pm$ 4.7	84.5 $\pm$ 7.2	73.9 $\pm$ 3.3
	FC-Max	85.7 $\pm$ 3.0	94.7 $\pm$ 0.9	67.4 $\pm$ 5.6	77.6 $\pm$ 4.3	82.9 $\pm$ 6.6	76.5 $\pm$ 1.4
	ACMIL	86.2 $\pm$ 6.3	93.6 $\pm$ 1.4	68.8 $\pm$ 2.4	78.6 $\pm$ 3.2	83.4 $\pm$ 7.6	75.8 $\pm$ 1.4
	eWSI 64	90.9 $\pm$ 3.9	95.7 $\pm$ 1.3	73.0 $\pm$ 5.6	82.1 $\pm$ 3.4	90.4 $\pm$ 5.2	80.7 $\pm$ 0.6

Table 2: Performance Comparison (% AUC) on TCGA and BRACS datasets.

Figure 2: Robustness of MILs to patch sampling for each initialization under frozen encoder settings.

\bigstar

denotes the PEFT+Lin-Max 3 performance under that setting. Experiments are conducted on Camelyon16, which is particularly sensitive to sampling rate due to smaller regions of interest.

Table 5

Method	$p_{t}$	iNet-Sup		iNet-SSL		Hist-SSL
Method	$p_{t}$	AUC	Acc.	AUC	Acc.	AUC	Acc.
ACMIL	0.50	78.3	72.7	90.4	79.8	96.7	90.7
eWSI 64	0.50	90.6	46.5	95.8	45.0	97.7	48.1
eWSI 64	0.98	90.6	82.9	95.8	88.4	97.7	92.2
eWSI 64 +FE	0.50	91.3	83.7	96.9	94.6	96.9	93.8

Table 6: Camelyon16 performance with bias correction.

p_{t}

: Probability threshold, BC: Manual bias correction, FE: Frozen encoder fine-tuning with dense sampling starting from eWSI encoder features. ACMIL: frozen encoder with

M=8192

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 4: An outline of eWSI. The input patches are randomly sampled and fed into the $\mathtt{LoRA}$ encoder. The encoded patches are then passed through the LinMax aggregator to yield the global WSI representation $\mathbf{z}$ .

Figure 5: Patch-predictions versus tumor locations for the 4 observed false negatives using eWSI iNet-SSL and $M=64$ . Yellow dots in the heatmaps indicate locations with high probability. The slide IDs are top left: $\mathtt{test\_038}$ , top right: $\mathtt{test\_011}$ , bottom left: $\mathtt{test\_013}$ , and bottom right: $\mathtt{test\_099}$ .

Figure 6: Training convergence under sparse and dense sampling. (a) MIL with

M{=}8192

, (b) MIL with

M{=}64

, and (c) eWSI with

M{=}64

Method	AUC %		Wall time (min)
Method	iBOT-B	Phikon	Patch	Training	Total
ACMIL (Frozen)	88.08 $\pm$ 0.7	99.4 $\pm$ 0.3	48.5	6.7	55.2
eWSI 64	91.61 $\pm$ 2.2	99.0 $\pm$ 0.3	-	45.8	45.8

Appendix

Table 7: Camelyon16 results under different encoder settings. ‘Patch’ denotes the time taken (under inference mode) to pre-compute patches of the training samples of the downstream dataset.

Figure 7: Iterative amplification and attenuation of features in the WSI using LinMax.

Figure 8: Effect of gate-like activation functions

Figure 9: Change of weights in magnitude (top) and direction (bottom) from iNet-SSL to Hist-SSL.

Figure 10: Effect of magnification vs. sampling rate $M$ on Camelyon16 performance. Experiments performing with a frozen encoder setting and ACMIL. horizontal axis is the sampling rate and vertical axis is the magnification

Figure 11: Sub-image representations and the pre-pooling activation maps for iNet-sup, iNet-sup+eWSI and Hist-SSL.

Explore similar work

Masked Feature Encoding for Large-Scale Whole Slide Image Representation
Oct 7, 2026Haoyu He, Basile Tessier-Cloutier, Yang Wang +1Whole-Slide Image Analysis Medical Image Representation Learning
Department of Computer Science and Software Engineering (CSSE), Concordia University, Montreal, Canada · McGill University, Montreal, Canada · Mila – Quebec AI Institute, Montreal, Canada
LRMIL: Efficient Low-Resolution Multiple Instance Learning via High-Resolution Knowledge Distillation for Whole Slide Image Classification
Jun 5, 2026Yonghan Shin, Won-Ki JeongHistopathology Image Classification Knowledge Distillation
Department of Computer Science and Engineering, Korea University, Seoul, Korea
Test-Time Instance Selection for Improved Whole Slide Image Analysis
Aug 14, 2026Quoc Anh Nguyen, Sunhong Park, Jin Tae KwakWhole-Slide Image Analysis Histopathology Image Classification
School of Electrical Engineering, Korea University, Seoul 02841, Korea

Oct 7, 2026J/K move · Enter open · S save

Masked Feature Encoding for Large-Scale Whole Slide Image Representation

Haoyu He, Basile Tessier-Cloutier, Yang Wang, Mahdi S. Hosseini

Department of Computer Science and Software Engineering (CSSE), Concordia University, Montreal, Canada · McGill University, Montreal, Canada · Mila – Quebec AI Institute, Montreal, Canada

Whole-Slide Image Analysis Medical Image Representation Learning Histopathology Image Classification Multiple Instance Learning Masked Image Modeling

Open paper AskPDF arXiv

Whole slide image (WSI) analysis in computational pathology follows a multiple instance learning (MIL) pipeline where patch embeddings are extracted independently and aggregated for slide-level prediction, but within-slide variance from staining, scanner, and local texture can overwhelm the discriminative signal. We propose Masked Feature Encoding for Multiple Instance Learning (MFE-MIL), a feature-space masking framework that trains a lightweight MLP adapter jointly with a window-based masked reconstruction branch and a MIL classification head. The two objectives are complementary. Classification guides the adapter to suppress within-slide patch variance, while window-based masked reconstruction provides an auxiliary regularizer for the adapted features without using patch coordinates, coordinate graphs, or segmentation preprocessing. The raster patch-extraction order is used only as a weak implicit prior. At inference, the decoder is removed, leaving only the adapter and MIL head. Across CAMELYON16/17, PANDA, and TCGA-BRCA with four diverse encoders, MFE-MIL improves ACC/F1 for nearly all tested aggregator-encoder settings and AUC in most, outperforms coordinate-based spatial methods (CAMIL), and achieves higher AUC than 2DMamba on three of four datasets (UNI). On five TCGA survival cohorts it improves the average concordance index for every aggregator tested, its most consistent gain. Code is available at https://github.com/AtlasAnalyticsLab/MFE-MIL.

Abstract

Explore similar work

Masked Feature Encoding for Large-Scale Whole Slide Image Representation

LRMIL: Efficient Low-Resolution Multiple Instance Learning via High-Resolution Knowledge Distillation for Whole Slide Image Classification

Test-Time Instance Selection for Improved Whole Slide Image Analysis