Unlocking the Potential of Digital Pathology: Novel Baselines for Compression

Alexander Muckenhuber; Constantin Ulrich; David Clunie; Jens Kleesiek; Klaus Maier-Hein; Marco Nolden; Maximilian Fischer; Michael Baumgartner; Michael G\"otz; Peter Neher

REVIEW 4 major objections 6 minor 3 references

Unlocking the Potential of Digital Pathology: Novel Baselines for Compression

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a task-agnostic metric, the cosine similarity between deep features of an original and a compressed whole-slide image, predicts downstream task performance well enough to standardize evaluation of lossy compression…

desk verdict Useful compression benchmark with a valuable uncompressed RMS dataset, but the headline task-agnostic metric is only qualitatively validated and the appendix undercuts its universal framing. read the letter →

arxiv 2412.13137 v1 pith:RQPHN6WZ submitted 2024-12-17 eess.IV cs.CV

Maximilian Fischer , Peter Neher , Peter Schüffler , Sebastian Ziegler , Shuhan Xiao , Robin Peretzke , David Clunie , Constantin Ulrich

show 8 more authors

Michael Baumgartner Alexander Muckenhuber Silvia Dias Almeida Michael Götz Jens Kleesiek Marco Nolden Rickmer Braren Klaus Maier-Hein

This is my paper · ORCID

classification eess.IVcs.CV

keywords wholeslideimageslossycompressionfeaturesimilaritydigitalpathologylearnedimageperceptualqualitydownstreamtaskperformancetask-agnosticmetric

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that the quality of a lossy compression scheme for whole-slide pathology images can be judged by a single task-agnostic number: the cosine similarity between deep features extracted from the original image and from its compressed version. If true, researchers and clinics could compare compression schemes without implementing and running every downstream task, such as classification, segmentation, or biomarker prediction, separately. The paper backs this by jointly benchmarking six codecs on perceptual metrics and two downstream tasks across four datasets, including a newly collected uncompressed rhabdomyosarcoma cohort. Its central empirical finding is that fine-tuned learned codecs win on their home domain but transfer poorly, while the feature-similarity ranking tracks actual downstream accuracy, with the stain-aware SQLC codec at the top.

What carries the argument

The load-bearing object is a task-agnostic feature-similarity metric: for each tile, features are extracted from the original and compressed image by a histopathology-pretrained ResNet18, and the cosine similarity between the two feature vectors is computed after flattening at several network depths. The metric operationalizes the idea that a compressed image is downstream-safe when its representation still points in the same direction in feature space as the original. It is the only component of the study that is proposed as a general standard, and the paper argues it tracks the ordering of six codecs on real classification and segmentation tasks.

What would settle it

Take any compression scheme not in the paper, compress the RMS test tiles, compute the proposed cosine similarity, and run a downstream task not used here, such as biomarker prediction or tissue classification with a different model architecture. If the ranking of codecs by feature similarity disagrees with the ranking by task accuracy, the general claim fails; the paper currently reports alignment only for a segmentation and a classification task, and only qualitatively.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that deep feature similarity is a reliable surrogate for downstream task performance on lossy compressed whole-slide images. The authors compute cosine similarity between embeddings of original and compressed tiles using a ResNet18 pretrained on histopathology, at six depths from early convolutional features to the final fully connected layer, and find that compression schemes with higher mean similarity are the same schemes that win classification and segmentation evaluations. The same benchmark shows that a codec fine-tuned on JPEG80 data for perceptual fidelity dominates perceptual metrics on JPEG80 data but loses that edge on uncompressed data, while the stain-aware SQLC codec, optimized for downstream performance, produces the highest feature similarity and the best downstream results. The paper therefore argues that feature similarity can replace per-task downstream evaluation as a general, standardized compression benchmark.

Load-bearing premise

The load-bearing premise is that the similarity of deep features between an original and a compressed image, measured by one pathology-trained feature extractor, predicts what any downstream model will need from that image; a second premise is that the RMS cohort really is uncompressed, which the paper asserts without verification.

Editorial extensions

If this is right

If feature similarity is a valid proxy, compression benchmarking no longer needs a separate implementation of each downstream task; one frozen feature extractor and a cosine distance suffice.
The benchmark ranking implies that WebP, a conventional codec, is the most practical near-term successor to JPEG for whole-slide storage because it is nearly as good as the best learned codecs but far faster.
Learned codecs that are fine-tuned on one compression artifact distribution should be expected to degrade when applied to data compressed differently, so deployment requires matching training and deployment compression settings.
The long decoding times of learned codecs, roughly 100 seconds per whole slide for SQLC at low bitrate, are a concrete barrier to clinical use independent of quality gains.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A testable extension the paper does not run: compress the same tiles with each codec at several bitrates, compute the proposed cosine similarity, and fit a monotone mapping to task accuracy for a third task such as mutation prediction; agreement would strengthen the claim that the metric is task-agnostic.
Because the metric depends on the feature extractor, adopting it would create a de facto standard extractor; different foundation models may rank codecs differently, especially if their training data already contained compression artifacts.
The appendix result where a large foundation model performed best on JPEG70-compressed input rather than on the original suggests that some feature extractors are adapted to compressed inputs; this could mean the metric measures match to the extractor's training distribution as much as information preserved by the codec.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Two things you should know. First, this is a useful benchmark paper: six lossy compression schemes, four datasets, perceptual and downstream tracks, plus a newly assembled uncompressed RMS dataset. Second, the headline novelty—a task-agnostic feature-similarity metric—is plausible but under-validated. The paper says the metric 'aligns very well' with downstream performance, but the evidence is a qualitative ordering at one bitrate, not a correlation, scatter, or test.

The strongest empirical contributions are the uncompressed RMS dataset and the finding that perceptual fine-tuning doesn't transfer: SPL2, trained on JPEG80 data, does poorly on uncompressed RMS, and neural codecs generalize poorly across compression schemes. The PanNuke segmentation and Camelyon16 classification experiments are reasonably structured, and the encoding/decoding time evaluation is a practical addition. The authors are also honest in the appendix, where they call the RMS classification experiments preliminary and list their own limitations.

The soft spots are real. The metric claim is the central contribution, and it rests on a single paragraph in §4.2. We don't get a single number linking cosine similarity to Dice or AUC. The metric is only evaluated at 0.5 bpp on a combined test set, so we can't tell whether the alignment holds across bitrates or tasks. The appendix gives a direct reason to be cautious about the 'task-agnostic' framing: the UNI foundation model gets its best AUC on JPEG70-compressed RMS data, beating the original uncompressed data. That doesn't falsify the specific Ciga-feature metric, since UNI uses different features, but it does undermine the assumption that distance-to-original is a universal proxy for downstream utility. There's also a reproducibility issue: the main text says 100 RMS subjects, the appendix says 96, and no code or evaluation scripts are provided. That's fixable but needs to be addressed.

Overall this is a solid empirical contribution with an overclaimed metric. The benchmark itself is worth having, and the negative result about neural codec generalization is credible. The metric needs quantitative validation before the central claim is accepted.

My recommendation: send it to peer review. A good referee can push for the correlation analysis and the dataset consistency check. If those get fixed, the paper becomes a useful reference for the pathology compression community.

Referee Report

4 major / 6 minor

Summary. The paper benchmarks six lossy compression schemes (JPEG, JPEG-XL, WebP, CAI, SPL2, SQLC) for whole-slide images in two tracks: perceptual quality (PSNR, MS-SSIM, LPIPS) and downstream tasks (nucleus segmentation with nnU-Net on PanNuke, and classification on Camelyon16). It introduces a task-agnostic evaluation metric computed as the cosine similarity between deep features of original and compressed images, using Ciga's ResNet18 pathology foundation model as the feature extractor. The authors claim this metric 'aligns very well' with actual downstream performance, allowing standardized evaluation without per-task testing. They also collect an 'initially uncompressed' RMS dataset for evaluating initial compression and report encode/decode times. The main contributions are the new metric, the broad comparative evaluation across perceptual and downstream criteria, and the uncompressed RMS dataset.

Significance. If the proposed task-agnostic metric were validated, it would provide a valuable, standardized tool for evaluating lossy compression schemes in digital pathology, reducing the need for expensive per-task downstream evaluation. The paper also provides a useful broad comparison of conventional and neural codecs, including a new uncompressed dataset and runtime analysis, which are strengths for reproducibility and practical uptake. However, the central validation of the metric is currently qualitative, and an appendix result directly contradicts the assumption that the original image is the upper baseline. The paper's strengths include the use of public datasets and checkpoints, transparent experimental design, and the inclusion of compute-time measurements.

major comments (4)

[Section 4.2, Figure 6] The central claim that feature cosine similarity 'aligns very well' with downstream performance is not quantitatively supported. The only evidence is a qualitative observation that SQLC has the highest mean similarity and also performed best in the classification and segmentation tasks, at a single bitrate (0.5 bpp) on a combined test set. No correlation coefficient, scatter plot, or statistical test links the cosine similarity values to Dice scores or classification accuracy, and no evaluation across multiple bitrates or tasks is presented. Please provide a quantitative comparison across schemes and bitrates, with uncertainty estimates, to substantiate the claim.
[Appendix, Table 1] The appendix's RMS classification experiments provide a direct counterexample to the metric's premise. The UNI foundation model achieves its best AUC (0.9575) with JPEG70 compression, outperforming the original uncompressed data (0.9383). This contradicts the assumption stated in Section 3.2 that the original image is the upper baseline and that images more similar to the original yield comparable downstream performance. Although the appendix uses UNI features rather than the Ciga features used in the proposed metric, it undermines the general claim that the metric is task-agnostic. The paper should either restrict the claim to the specific feature extractor used or explain why this result does not violate the premise.
[Section 2.2, RMS dataset] The main text states that 'we collected 100 randomly selected subjects' and extracted 1000 tiles per subject, resulting in 100,000 uncompressed tiles, while the Appendix states that the RMS dataset 'consists of 96 subjects' (52 Embryonal and 44 Alveolar). This inconsistency affects the reproducibility of the uncompressed dataset and the downstream experiments. Please clarify the correct dataset size and reconcile the numbers.
[Section 2.2, RMS uncompressed claim] The paper asserts that the RMS data 'has not been previously lossy compressed' and is 'ideally suited to evaluate compression schemes for the initial compression,' but provides no verification that the WSI container or scanner pipeline introduced no lossy steps. Since both the perceptual evaluation and the proposed metric treat the original as the gold standard, this assumption is load-bearing. Please provide evidence from container metadata or scanner settings, or at least discuss the risk that prior lossy processing could bias the results.

minor comments (6)

[Appendix, References] In the appendix, reference [6] is cited for the UNI foundation model, but the appendix reference list identifies [6] as OpenSlide; UNI is reference [8]. The citation should be corrected.
[Section 3.2, Eq. (1)] The subscript in the cosine similarity notation appears garbled as 's!'#'; please use a readable symbol such as s_cos to denote the metric.
[Figure 7 caption] The caption reads 'we the logarithmic time' and appears to be missing a verb; it should read 'we show the logarithmic time.'
[Appendix] The appendix is not referenced anywhere in the main text. Add a cross-reference in Section 4 or 5 so readers are aware of the additional RMS classification experiments.
[Section 5, Discussion] There is a typo in the phrase 'WSI comression' in the discussion; it should be 'WSI compression.'
[Section 5, Discussion] The statement 'we do not consider approaches with auto encoders, that solely perform dimensionality reduction' is confusing because the neural codecs in the study are autoencoders; clarify that the distinction is about dimensionality-reduction-only methods without quantization and bitstream coding.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the proposed feature-similarity metric is independent of downstream-task scores and is not fitted to the outcomes it is claimed to predict.

full rationale

The central metric is defined in Section 3.2 by Eq. (1) as the cosine similarity between Ciga ResNet18 features of the original and compressed images. No downstream accuracy value enters this definition, and no weighting or threshold is fitted to task performance; the metric is therefore not a fitted predictor of the tasks it is compared against. The validation in Section 4.2 compares the metric's ranking with independently measured Dice scores from the paper's own nnU-Net segmentation experiment and with the classification result cited from [26]. The only self-referential elements are the inheritance of the JPEG80 test split from [25] and the use of the same-authors' SQLC result for one of the two validation tasks; these are empirical anchors rather than definitional identities. The appendix's UNI/JPEG70 result (compression improving AUC over the original) challenges the assumption that the original is the upper baseline, but that is a validity concern, not a circularity. No step in the derivation reduces by construction to its own input.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical or conceptual entities. Its central contribution is a new evaluation metric, which sits on domain assumptions about the foundation model and the uncompressed status of the RMS data rather than on new components.

free parameters (1)

Evaluation bitrates = 0.5 bpp (task-agnostic metric), 0.7 bpp (segmentation visual), 0 to 1.75 bpp (perceptual curves)
Chosen by hand as comparison points; not fitted to any target. They affect the reported rankings but are not free parameters in a derivation.

assumptions (4)

domain assumption The RMS dataset is uncompressed (no previous lossy compression)
Section 2.2 states the rhabdomyosarcoma images 'have not been previously lossy compressed', which is the basis for calling this an unbiased initial-compression evaluation. If the scanner or DICOM conversion applied lossy steps, the baseline is compromised.
domain assumption Features from Ciga's ResNet18 foundation model are a valid proxy for downstream task information
Section 3.2 uses this model as the feature extractor for the task-agnostic metric. The validity of the metric depends on this premise.
domain assumption Cosine similarity in the feature space reflects downstream performance
The entire proposed metric rests on this assumption, stated in Section 3.2. It is not derived from theory; it is asserted and only qualitatively illustrated.
domain assumption The JPEG80 test split from [25] is representative for recompression evaluation
Section 2.2 reuses the test split from the authors' prior work [25]; if this split is biased, the perceptual recompression rankings could be skewed.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Unlocking the Potential of Digital Pathology: Novel Baselines for Compression." pith.science (2026). https://pith.science/paper/RQPHN6WZ

@misc{pith2026241213137,
  author       = {Pith},
  title        = {Pith review of: Unlocking the Potential of Digital Pathology: Novel Baselines for Compression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RQPHN6WZ}},
  note         = {Machine review of arXiv:2412.13137}
}

read the original abstract

Digital pathology offers a groundbreaking opportunity to transform clinical practice in histopathological image analysis, yet faces a significant hurdle: the substantial file sizes of pathological Whole Slide Images (WSI). While current digital pathology solutions rely on lossy JPEG compression to address this issue, lossy compression can introduce color and texture disparities, potentially impacting clinical decision-making. While prior research addresses perceptual image quality and downstream performance independently of each other, we jointly evaluate compression schemes for perceptual and downstream task quality on four different datasets. In addition, we collect an initially uncompressed dataset for an unbiased perceptual evaluation of compression schemes. Our results show that deep learning models fine-tuned for perceptual quality outperform conventional compression schemes like JPEG-XL or WebP for further compression of WSI. However, they exhibit a significant bias towards the compression artifacts present in the training data and struggle to generalize across various compression schemes. We introduce a novel evaluation metric based on feature similarity between original files and compressed files that aligns very well with the actual downstream performance on the compressed WSI. Our metric allows for a general and standardized evaluation of lossy compression schemes and mitigates the requirement to independently assess different downstream tasks. Our study provides novel insights for the assessment of lossy compression schemes for WSI and encourages a unified evaluation of lossy compression schemes to accelerate the clinical uptake of digital pathology.

Figures

Figures reproduced from arXiv: 2412.13137 by the authors.

**Figure 1.** Overview of our approach. We provide a full overview of the performanceof linear compression schemes as well as for m [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

**Figure 2.** Dataset BreaKHis Colon1 Colon2 Source [34] [35] [36] Tissue Breast Colon Colon Images 400 100 2 Sample Size 700x460 768x768 5000x5000 Tile Size 224x224 224x224 224x224 Tiles 2000 2000 1000 [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

**Figure 2.** Overview of the RMS dataset. The dataset was acquired without lossy [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

Figures from the paper (6 more)

**Figure 3.** Figure 3: Rate-distortion and -perception curves on the test set for various metrics. Arrows (↑), (↓) indicate if high or low scores are preferable. The bits-per-pixel (bpp) determines the level of compression. In our experiments, we evaluate both perspectives: perceptual image …

**Figure 4.** Figure 4: Image quality metrics for compressing uncompressed [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]

**Figure 5.** Figure 5: Visualization of predicted segmentations along the numeric segmentation accuracy. [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]

**Figure 6.** Figure 6: Averaged cosine similarity between the original images and the compressed counterpart. Each violin refers to a differ [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]

**Figure 7.** Figure 7: De- vs. encoding times for the compared compression schemes on the test set. The Figure shows the measured times for various compression ratios. 5 Discussion Deep learning-based lossy compression schemes as well as more recent conventional compression schemes outperfor…

**Figure 1.** Figure 1: Figure1: Results of our classification experiments on the RMS data. [PITH_FULL_IMAGE:figures/full_fig_p022_1.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

3 extracted references · 1 canonical work pages

[3]

Expert annotations of the tissue types for the RMS-Mutation-Prediction microscopy images

Bridge, C. (2024) “Expert annotations of the tissue types for the RMS-Mutation-Prediction microscopy images”. Zenodo. doi: 10.5281/zenodo.10462858. [4] Chen, Y. et al. (2020) Quantitative Assessment of the Effects of Compression on Deep Learning in Digital Pathology Image Analysis. JCO Clin Cancer Inform 4, 221-233. DOI:10.1200/CCI.19.00068 [5] Ilse, M. e...

arXiv 2024
[15]

Learned Image Compression for HE-stained Histopathological Images via Stain Deconvolution

C. L´opez, “Effects of image compression on automatic count of immunohistochemically stained nuclei in digital images,” in Journal of the American Medical Informatics Association, vol. 15, 2008. 16. S. Doyle, “Evaluation of effects of JPEG2000 compression on a computer-aided detection system for prostate cancer on digitized histopathology,” in 2010 IEEE I...

work page Pith review arXiv 2008
[40]

Pannuke: An open pan-cancer histology dataset for nuclei instance segmentation and classification,

G. Jevgenij, “Pannuke: An open pan-cancer histology dataset for nuclei instance segmentation and classification,” in European Congress on Digital Pathology, 2019. 41. Y. Yang, “An introduction to neural data compression,” Foundations and Trends® in Computer Graphics and Vision, 2023. 42. Z. Wang, “Multiscale structural similarity for image quality assessm...

arXiv 2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [3]

Expert annotations of the tissue types for the RMS-Mutation-Prediction microscopy images

Bridge, C. (2024) “Expert annotations of the tissue types for the RMS-Mutation-Prediction microscopy images”. Zenodo. doi: 10.5281/zenodo.10462858. [4] Chen, Y. et al. (2020) Quantitative Assessment of the Effects of Compression on Deep Learning in Digital Pathology Image Analysis. JCO Clin Cancer Inform 4, 221-233. DOI:10.1200/CCI.19.00068 [5] Ilse, M. e...

arXiv 2024

[2] [15]

Learned Image Compression for HE-stained Histopathological Images via Stain Deconvolution

C. L´opez, “Effects of image compression on automatic count of immunohistochemically stained nuclei in digital images,” in Journal of the American Medical Informatics Association, vol. 15, 2008. 16. S. Doyle, “Evaluation of effects of JPEG2000 compression on a computer-aided detection system for prostate cancer on digitized histopathology,” in 2010 IEEE I...

work page Pith review arXiv 2008

[3] [40]

Pannuke: An open pan-cancer histology dataset for nuclei instance segmentation and classification,

G. Jevgenij, “Pannuke: An open pan-cancer histology dataset for nuclei instance segmentation and classification,” in European Congress on Digital Pathology, 2019. 41. Y. Yang, “An introduction to neural data compression,” Foundations and Trends® in Computer Graphics and Vision, 2023. 42. Z. Wang, “Multiscale structural similarity for image quality assessm...

arXiv 2024