Pith. sign in

REVIEW 3 major objections 6 minor 19 references

Deep Learning for Visual Recognition of Environmental Enteropathy and Celiac Disease

T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A ResNet50 trained on color-normalized patches of H&E duodenal biopsies can separate Celiac Disease, Environmental Enteropathy, and Normal tissue, reporting 92.86% whole-slide accuracy and macro-average AUC 0.99.

desk verdict Plausible transfer-learning result for a hard differential diagnosis, but the perfect EE scores smell like site confound; needs a within-site or patient-level check before the accuracy is believable. read the letter →

arxiv 1908.03272 v1 pith:2BPE5UMG submitted 2019-08-08 q-bio.QM cs.CVeess.IV

Aman Shrivastava , Karan Kant , Saurav Sengupta , Sung-Jun Kang , Marium Khan , Asad Ali , Sean R. Moore , Beatrice C. Amadi
show 3 more authors
Paul Kelly Donald E. Brown Sana Syed
This is my paper · ORCID
classification q-bio.QMcs.CVeess.IV
keywords deeplearningduodenalbiopsyceliacdiseaseenvironmentalenteropathywholeslideimagestransferstainnormalizationGrad-CAM
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 a deep convolutional network can separate three histologically similar duodenal biopsy categories—Celiac Disease, Environmental Enteropathy, and Normal—at an accuracy high enough to be a plausible assistive diagnostic. It reports whole-slide accuracy of 92.86%, with perfect precision and recall for Environmental Enteropathy and a macro-average AUC of 0.99. This matters because EE and Celiac Disease share overlapping features that can confuse human pathologists, so a reproducible computational read could help standardize diagnosis. The paper also argues that the model's attention maps focus on pathology-like structures rather than obvious image artifacts.

What carries the argument

The load-bearing object is a ResNet50 convolutional network pre-trained on a large natural-image dataset, fine-tuned discriminatively with cyclic cosine annealing and test-time augmentation, and trained on 256-by-256 resized patches extracted from whole-slide images. A structure-preserving stain-normalization step maps all slides to a common color appearance, intended to strip scanner and laboratory variation so the network learns tissue features instead of color differences. Patch-level predictions are majority-aggregated to whole-slide labels; Grad-CAM and filter visualizations are then used to inspect which tissue regions and intermediate features drive the classification.

What would settle it

Take the same three-class task but train and test on slides drawn from every site, or from one site scanned on multiple scanners, and see whether accuracy stays above 90%. A second check: train a classifier on the stain-normalized patches' color histograms alone; if it separates classes nearly as well, the visual features are dominated by residual site artifacts.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that transfer-learned ResNet50 models applied to color-normalized patches of H&E duodenal biopsies classify the three classes at the whole-slide level with 92.86% accuracy and a macro-average AUC of 0.99. Patch-level predictions are aggregated to whole-slide labels; Environmental Enteropathy is recognized with perfect precision and recall, while Celiac Disease and Normal show more confusion with each other. The paper reports that fusing features from two magnification levels gave little benefit over single-scale models, and that visual explanation methods suggest the network attends to nuclei, cell boundaries, and tissue-background patterns rather than pure color differences.

Load-bearing premise

The entire result depends on the assumption that the stain-normalization step erases the color and scanner differences between the sites where the biopsies were collected, so that the network is learning disease biology rather than which hospital produced the slide.

Editorial extensions

If this is right

  • If the reported accuracy holds in clinical settings, automated triage of duodenal biopsies could flag likely Celiac Disease or Environmental Enteropathy cases before expert review.
  • The near-perfect recognition of Environmental Enteropathy suggests screening for EE in populations with growth faltering could be automated from routine biopsy slides.
  • The finding that multi-zoom feature fusion adds little performance benefit implies single-magnification patch models may be sufficient for similar histopathology classification tasks.
  • The interpretability evidence indicates the model's decisions can be aligned with the features pathologists look for, supporting use as a second-reader tool.
  • The same transfer-learning pipeline could plausibly be applied to other enteropathies with overlapping histology.

Reading between the lines

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

  • My inference: because all Environmental Enteropathy slides came from two clinical sites while Celiac Disease and Normal slides came from a third site, the 0.99 AUC may partly encode site or scanner identity rather than disease biology; the paper does not test this directly.
  • A cross-site validation set with EE, Celiac Disease, and Normal slides drawn from every site would settle whether stain normalization fully removes the site confound; until then, site identity remains a plausible alternative explanation for EE's perfect recall.
  • The choice of the stain-normalization target image drawn from the EE set is a free parameter; my expectation is that changing the target could shift the decision boundary, and sensitivity to that choice should be reported.
  • If the method transfers to new sites, an obvious next step is weakly supervised whole-slide classification that skips explicit patch aggregation, which would remove the need for the multi-zoom fusion the paper found unhelpful.
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.

Referee Report

3 major / 6 minor

Summary. The manuscript presents a deep learning pipeline to classify duodenal biopsy whole-slide images into Celiac Disease, Environmental Enteropathy, and Normal tissue. Using a ResNet50 pretrained on ImageNet, patch extraction at two magnifications, Vahadane stain normalization, data augmentation, and test-time augmentation, the authors report patch-level accuracies of 88.89% (1000x1000) and 86.82% (2000x2000), and a whole-slide accuracy of 92.86% on 98 held-out slides with macro-AUC 0.99. The paper also presents Grad-CAM and filter visualizations to support model interpretability.

Significance. If the reported results are robust, the approach could offer a reproducible assistive tool for distinguishing histologically overlapping enteropathies, particularly for Environmental Enteropathy, for which computational pathology support is scarce. The paper demonstrates a reasonable transfer-learning setup and includes interpretability analyses. However, the central claim is currently limited by the confounding of disease class with acquisition site, the lack of a patient-level split, and the small EE test set; these limitations prevent the study from establishing that the classifier generalizes on disease biology rather than site-specific artifacts.

major comments (3)
  1. [Section II and Section III-B] The dataset is built so that disease class is completely confounded with acquisition site: all EE WSIs originate from Aga Khan University Hospital and University Teaching Hospital in Zambia, while all CD and Normal WSIs originate from the University of Virginia. Stain normalization to an empirically chosen target image from the EE dataset (Section III-B) is invoked to remove color differences, but no evidence is given that it removes scanner- and protocol-specific batch effects. The fact that unnormalized images yield 'erroneous results' based on color difference indicates the kind of signal the model can exploit; without a quantitative demonstration that normalization eliminates site-discriminative information (e.g., a site-classification control experiment), the perfect EE precision/recall in Table II can be explained by acquisition site. This must be addressed with a site-holdout evaluation or an explicit normalization-validation experiment.
  2. [Section II / IV-A] The paper does not describe whether the 367 training and 98 test slides are disjoint at the patient level. With 150 children contributing 465 WSIs, multiple WSIs per child are present; if slides from the same patient appear in both sets, patch-level leakage can inflate the reported accuracy. The authors should confirm that the split is by patient and, if it is not, repeat the evaluation with patient-level cross-validation.
  3. [Table II / Section IV-A] The EE test set contains only 11 whole-slide images, and the paper reports no confidence intervals or statistical tests for the 92.86% accuracy or macro-AUC of 0.99. With n=11, the 1.00 precision/recall for EE is not a stable estimate; a single error changes the metric by roughly 9 percentage points. The abstract's 'over 90%' claim should be qualified accordingly.
minor comments (6)
  1. [Section IV-A] The phrase 'unlabeled 98 slides' is misleading because Table II provides the true class support counts; rename this to 'held-out test slides'.
  2. [Section IV-B] The text states 'Figure 3 displays a few example Grad-CAM outputs', but Figure 3 is the ROC curves; the Grad-CAM images appear to be in Figure 2. Please correct the cross-references.
  3. [Section III-C] Test-time augmentation is described as random during prediction; for reproducible and stable predictions, use a fixed set of augmentations and average the outputs.
  4. [Section III-E] The multi-zoom approach is described but no quantitative results are reported; either present the comparison or remove the section.
  5. [Section II] The relationship between patients, WSIs, and slides ('465 high resolution whole slide images taken from 150 H&E duodenal biopsy slides') is unclear; clarify the unit of analysis and state whether multiple WSIs from the same patient are treated as independent.
  6. [References] Reference [6] is a prior abstract by the same group; the paper should explicitly state the new contribution beyond that abstract.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the accuracy claim is an empirical held-out evaluation, and the self-citation and stain-normalization choice are not construction-equivalent inputs.

full rationale

The paper is an empirical machine-learning study, not a derivation. The chain is: labeled slides are split into patches, stain normalization is applied, a ResNet50 with transfer learning is trained on patches from 367 slides, patch predictions are aggregated, and the resulting model is evaluated on 98 previously unseen slides (Table II). No fitted parameter or training label is reused as the reported outcome, and the test labels are not used in training. The stain-normalization target image is chosen from the EE dataset, but it is a preprocessing reference, not a learned parameter, and the reported accuracy is not defined in terms of that target; this raises a possible color-bias concern but not a circular reduction. The only overlapping-author citation (Ref. [6], a prior Gastroenterology abstract) appears in the introduction as background on prior CNN use in duodenal biopsies and is not load-bearing for the paper's accuracy claim. The site/color confounding between EE slides from Pakistan/Zambia and CD/Normal slides from UVA is a genuine external-validity and correctness risk, but confounding is not circularity: even a scanner-discrimination model would still have its accuracy estimated on held-out data. Under the stated circularity definitions, the central claim is self-contained, so the score is 0.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim depends on the validity of clinical/endoscopic labels, the absence of patient leakage, the effectiveness of stain normalization in removing site confounds, and the transferability of ImageNet features. These are all domain assumptions; no new physical entities are introduced.

free parameters (5)
  • stain normalization target image = not specified (chosen from EE dataset)
    Used to map all images' colors; chosen ad hoc from one class, could bias normalization toward EE appearance.
  • upsampling factor = not specified (reported as 'appropriate factor')
    Used to balance class patch counts; affects training distribution and could leak augmented copies into test if done before splitting.
  • learning rate ratios = 1/9 and 1/3 of final layer rate
    Discriminative fine-tuning schedule chosen from Howard and Ruder; affects optimization trajectory.
  • number of epochs = 10
    Training length; no early stopping or model selection details provided.
  • batch size = 32
    Standard choice; affects training dynamics.
assumptions (5)
  • domain assumption Ground truth labels (CD, EE, Normal) are correct per clinical and endoscopic diagnosis.
    Section II assigns labels based on endoscopy and clinical context, without independent histopathologic confirmation or inter-observer agreement.
  • domain assumption Patches from the same slide are exchangeable and training and test slides do not share patients.
    The paper does not state a patient-level split; if the same patient contributes to both training and test, accuracy is inflated.
  • domain assumption Stain normalization removes inter-site color variations while preserving pathological features.
    Section III-B relies on Vahadane et al. with a single target image; no validation that disease signal is preserved.
  • domain assumption ImageNet transfer learning provides useful features for histopathology.
    Section III-D uses ImageNet pretraining; a common but unproven assumption for this domain.
  • domain assumption The three classes are mutually exclusive and each slide has one label.
    Needed for the classification setup; mixed pathology or uncertain diagnoses are not discussed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning for Visual Recognition of Environmental Enteropathy and Celiac Disease." pith.science (2026). https://pith.science/paper/2BPE5UMG

@misc{pith2026190803272,
  author       = {Pith},
  title        = {Pith review of: Deep Learning for Visual Recognition of Environmental Enteropathy and Celiac Disease},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2BPE5UMG}},
  note         = {Machine review of arXiv:1908.03272}
}
read the original abstract

Physicians use biopsies to distinguish between different but histologically similar enteropathies. The range of syndromes and pathologies that could cause different gastrointestinal conditions makes this a difficult problem. Recently, deep learning has been used successfully in helping diagnose cancerous tissues in histopathological images. These successes motivated the research presented in this paper, which describes a deep learning approach that distinguishes between Celiac Disease (CD) and Environmental Enteropathy (EE) and normal tissue from digitized duodenal biopsies. Experimental results show accuracies of over 90% for this approach. We also look into interpreting the neural network model using Gradient-weighted Class Activation Mappings and filter activations on input images to understand the visual explanations for the decisions made by the model.

Figures

Figures reproduced from arXiv: 1908.03272 by the authors.

Figure 1
Figure 1. Stain Normalization on patches. Images in the top row highlight the difference in stain color before normalization and the bottom row shows [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Gradient-weighted Class Activation Maps for diseased tissue patches. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. ROC curves on unsampled dataset TABLE II TEST SET ACCURACY Class precision recall f1-score support Celiac 0.98 0.86 0.92 51 EE 1.00 1.00 1.00 11 Normal 0.83 0.97 0.90 36 avg / total 0.93 0.92 0.92 98 aWSI level accuracy. of utmost importance that we utilize methods to explain their classification result. Furthermore, visualizing activation areas allows for domain experts to corroborate the model results with incumbe… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Layer activation maps extracted for the given input image. [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 18 canonical work pages

  1. [6]

    S. Syed et al., ”195 - Convolutional Neural Networks Image Anal- ysis of Duodenal Biopsies Robustly Distinguishes Environmental Enteropathy from Healthy Controls and Identifies Secretory Cell Lineages as High Activation Locations”, Gastroenterology, vol. 154, no. 6, p. S-52, 2018

    work page 2018

  2. [1]

    Geboes and G

    K. Geboes and G. Y . Lauwers, ”Gastrointestinal Pathology,” Archives of Pathology & Laboratory Medicine, vol. 134, no. 6, pp. 812-814, June 2010

    work page 2010

  3. [2]

    R. E. Nakhleh, ”Error Reduction in Surgical Pathology,” Archives of Pathology & Laboratory Medicine, vol. 130, no. 5, pp. 630-632, May 2006

    work page 2006

  4. [3]

    S. S. Raab, D. M. Grzybicki, J. E. Janosky, R. J. Zarbo, F. A. Meier, C. Jensen and S. J. Geyer, ”Clinical impact and frequency of anatomic pathology errors in cancer diagnoses.,” Cancer, vol. 104, no. 10, p. 22052213, 2005

    work page 2005

  5. [4]

    M. N. Gurcan, L. E. Boucheron, A. Can, A. Madabhushi, N. M. Rajpoot and B. Yener., ”Histopathological image analysis a review.,” IEEE Reviews in Biomedical Engineering, pp. 147-171, 2009

    work page 2009

  6. [5]

    Sharma and A

    G. Sharma and A. Carter, ”Artificial Intelligence and the Pathologist, Future Frenemies?,” Archives of Pathology & Laboratory Medicine, vol. 141, pp. 622-623, May 2017

    work page 2017

  7. [7]

    Cruz-Roa, A

    A. Cruz-Roa, A. Basavanhally, F. Gonzlez, H. Gilmore, M. Feldman, S. Ganesan, N. Shih, J. Tomaszewski and A. Madabhushi, ”Automatic detection of invasive ductal carcinoma in whole slide images with Convolutional Neural Networks,” in SPIE Medical Imaging 2014: Digital Pathology, San Diego, California, United States, 2014

    work page 2014

  8. [8]

    ”Convolutional neural network architectures for the automated diagnosis of celiac disease.” International Workshop on Computer-Assisted and Robotic Endoscopy

    Wimmer, Georg, et al. ”Convolutional neural network architectures for the automated diagnosis of celiac disease.” International Workshop on Computer-Assisted and Robotic Endoscopy. Springer, Cham, 2016

    work page 2016

Show all 19 references
  1. [9]

    W., Wei, J

    Wei, J. W., Wei, J. W., Jackson, C. R., Ren, B., Suriawinata, A. A., Hassanpour, S., ”Automated detection of celiac disease on duodenal biopsy slides: a deep learning approach” arXiv:1901.11447 [cs.CV] Jan 2019

    1901 arXiv

  2. [10]

    C. L. Jansson-Knodell, I. A. Hujoel, A. Rubio-Tapia and J. A. Murray, ”Not All That Flattens Villi Is Celiac Disease: A Review of Enteropathies,” Mayo Clinic Proceedings, pp. 509-517, April 2018

    2018

  3. [11]

    Krizhevsky, I

    A. Krizhevsky, I. Sutskever and G. E. Hinton, ”ImageNet Classification with Deep Convolutional Neural Networks,” in Advances in Neural Information Processing Systems 25 (NIPS 2012), 2012

    2012

  4. [12]

    Vahadane, T

    A. Vahadane, T. Peng, A. Sethi, S. Albarqouni, L. Wang, M. Baust, K. Steiger, A. M. Schlitter, I. Esposito and N. Navab, ”Structure- Preserving Color Normalization and Sparse Stain Separation for Histological Images,” IEEE Transactions on Medical Imaging, vol. 35, no. 8, pp. 1...

    1962

  5. [13]

    B. E. Bejnordi, M. Veta, P. Johannes van Diest, B. van Ginneken, N. Karssemeijer, G. Litjens and J. A.W. M. van der Laak, ”Diagnostic Assessment of Deep Learning Algorithms for Detection of Lymph Node Metastases in Women With Breast Cancer,” Journal of the American Medical Ass...

    2017

  6. [14]

    S. J. Pan and Q. Yang, ”A Survey on Transfer Learning,” IEEE Transactions on Knowledge and Data Engineering, vol. 22, no. 10, pp. 1345-1359, 2010

    2010

  7. [15]

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li and L. Fei-Fei, ImageNet: A Large-Scale Hierarchical Image Database. IEEE Computer Vision and Pattern Recognition (CVPR), 2009

    2009

  8. [16]

    Howard and S

    J. Howard and S. Ruder, ”Universal Language Model Fine-tuning for Text Classification,” in Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Long Papers), Melbourne, 2018

    2018

  9. [17]

    Loshchilov., F

    I. Loshchilov., F. Hutter, SGDR: Stochastic Gradient Descent with Warm Restarts, arXiv:1608.03983 [cs.LG], May 2017

    2017 arXiv

  10. [18]

    R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh and D. Batra, ”GradCAM: Visual Explanations from Deep Networks via Gradient-Based Localization,” in IEEE International Conference on Computer Vision (ICCV), Venice, 2017

    2017

  11. [19]

    M. D. Zeiler and R. Fergus, ”Visualizing and understanding convo- lutional networks.,” in European conference on computer vision. , Springer, Cham, 2014

    2014

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.