Evaluation and optimization of deep learning models for enhanced detection of brain cancer using transmission optical microscopy of thin brain tissue samples

Mohnish Sao; Mousa Alrubayan; Prabhakar Pradhan

REVIEW 4 major objections 4 minor 19 references

Evaluation and optimization of deep learning models for enhanced detection of brain cancer using transmission optical microscopy of thin brain tissue samples

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read This paper argues that DenseNet-121, a convolutional network that reuses features through dense connections, outperforms ResNet-50 for detecting brain cancer in bright-field transmission microscopy images of thin tissue slices, reaching…

desk verdict A credible but overinterpreted two-model benchmark on a private brain TMA dataset; the numbers are internally consistent, but the dense-connectivity conclusion outruns the single-dataset, single-split evidence. read the letter →

arxiv 2505.11735 v1 pith:SHOFNS3R submitted 2025-05-16 physics.med-ph physics.bio-phphysics.optics

Mohnish Sao , Mousa Alrubayan , Prabhakar Pradhan This is my paper · ORCID

classification physics.med-phphysics.bio-phphysics.optics

keywords braincancerdetectionbright-fieldmicroscopydeepconvolutionalneuralnetworksDenseNet-121ResNet-50transferlearningtissuemicroarraymedicalimageclassification

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

This paper argues that DenseNet-121, a convolutional network whose layers reuse features through dense connections, is better suited than ResNet-50 for detecting brain cancer in bright-field transmission microscopy images of thin tissue slices. On a curated set of 2,931 images from tissue microarrays, DenseNet-121 reaches 88.35% test accuracy and a 0.9116 F1-score, against 82.12% and 0.8563 for ResNet-50, a gap the authors report as statistically significant (p=0.004, McNemar test). The authors read this as evidence that dense connectivity generalizes better on limited medical datasets, and they present their two-stage transfer-learning pipeline as a reusable blueprint for clinical bright-field workflows.

What carries the argument

The load-bearing object is DenseNet-121's dense connectivity: within each dense block, every layer receives the concatenated feature maps of all preceding layers, promoting feature reuse and gradient flow, and this is compared against ResNet-50's residual identity skip connections. The argument also depends on the two-stage transfer-learning protocol: first train a new classifier head on frozen ImageNet-pretrained features, then unfreeze the last two blocks and fine-tune with data augmentation, dropout, weight decay, learning-rate decay on plateau, and early stopping. This protocol is what the authors say lets both models converge cleanly on a 2,931-image dataset and lets the architectural difference show up as a statistically significant accuracy gap.

What would settle it

Run the identical DenseNet-121 versus ResNet-50 protocol on images where the cancer labels are randomly permuted across slices from the same slide, or on an independent multi-center TMA set with the same two-stage training; if the 6.23% accuracy gap and the high precision disappear or shrink to noise, the claim that dense connectivity is systematically better on limited brain-tissue datasets is not supported.

Watch

Extended reading notes

Core claim

The central claim is a head-to-head comparison: under identical two-stage transfer learning (frozen feature extractor trained for 10 epochs, then fine-tuning the last convolutional blocks for 20 epochs with augmentation, early stopping, and dropout), DenseNet-121 outperforms ResNet-50 on a held-out test set of 498 brain-tissue images. The reported numbers are 88.35% accuracy (95% CI ±1.28%), precision 0.9614, recall 0.8667, and F1 0.9116 for DenseNet-121, versus 82.12% accuracy (±1.75%), precision 0.9035, recall 0.8142, and F1 0.8563 for ResNet-50. The confusion matrix shows 299 true positives, 141 true negatives, 46 false negatives, and 12 false positives. The authors conclude that dense connectivity—connecting each layer to every other layer within dense blocks—improves feature propagation and gradient flow, and that this advantage is especially visible when training data are scarce.

Load-bearing premise

The whole comparison rests on the claim that the bright-field transmission intensity is proportional to tissue mass density, so that cancer-related structural changes produce a learnable optical signature in the images; if that link fails, or if the slice-level labels or the microscope slides carry batch artifacts, the reported accuracy and the DenseNet-versus-ResNet gap would not generalize beyond this dataset.

Editorial extensions

If this is right

DenseNet-121 could serve as an automated pre-screening tool for brain cancer on bright-field biopsy slides, reducing manual review time.
The reported low false-positive rate (7.9%) implies fewer unnecessary follow-up procedures if such a model is used as a filter.
Architecture choice matters more on small medical datasets: dense feature reuse may be preferable to residual shortcuts when labeled images number only in the thousands.
The two-stage transfer-learning recipe (frozen head, then selective fine-tuning) can be carried over directly to other tissue types or stain protocols.
The statistically significant gap (p=0.004) supports prioritizing DenseNet-style connectivity in clinical AI pipelines for histopathology.

Reading between the lines

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

The paper's architectural conclusion is drawn from a single dataset; on larger datasets or different tissue types the DenseNet advantage might shrink, since dense connections add memory and compute cost.
The claimed proportionality between transmission intensity and mass density could be tested directly by correlating model confidence or saliency maps with independent measurements such as quantitative phase or mass-density imaging.
A simple extension would be to switch from binary cancer detection to ordinal tumor grading; dense connectivity is naturally compatible with ordinal losses and the paper cites that direction.
One untested risk is slide-level batch effects: a leave-one-slide-out split, rather than the subject-level split used here, would reveal whether the model learns biology or slide-specific artifacts.

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

Quick take: this is a straightforward comparison of two off-the-shelf CNNs on a new private bright-field brain TMA dataset. The reported metrics are internally consistent, and splitting at the patient level is good practice. But the paper's broader claim—that DenseNet-121's dense connectivity is the reason it wins on limited medical data—is not supported by the design, and the slide-level batch confounder is not addressed.

What is actually new here is the specific benchmark: 498 test slices from a commercial brain TMA, with DenseNet-121 at 88.35% accuracy and ResNet-50 at 82.12%, p=0.004 by McNemar. The two-stage transfer learning protocol is described in enough detail to reproduce, the confusion matrix matches the stated metrics, and bootstrap CIs are given. That is a fair amount of care for a small empirical paper, and credit is due for the patient-level split.

The soft spots are real though. The biggest one is the one the stress-test flags: all cores on a TMA slide share the same staining, sectioning, and illumination, but the paper never reports how slides are distributed across train/validation/test or whether cancer and control cores are balanced within slides. If the classes are concentrated on different slides, the classifier can be reading slide-level brightness or stain variation rather than tumor morphology. That would inflate both the absolute accuracy and the DenseNet advantage, and it would gut the 'dense connectivity generalizes better' conclusion. This is not a nitpick; it is a testable design question that should have been answered in Section 2.1.

Other issues are more minor: no multi-run variance, no external validation, no code or data release. The writing is sloppy in places—Section 2.1's intensity-to-mass-density sentence is garbled and repeated, and references [10], [11], and [16] are misapplied (one is a Pseudomonas paper, not brain tissue). The 'Limitations' paragraph is an incomplete sentence. None of these change the numbers, but they signal the manuscript needs another editing pass and a careful citation check.

Bottom line: this is a reasonable empirical data point for someone building a benchmark in small-sample bright-field histology, and it deserves a serious referee if the authors can add a slide-level batch analysis and temper the architectural claim. Without those, it is a single-dataset result, not a general lesson.

Referee Report

4 major / 4 minor

Summary. The manuscript evaluates ResNet-50 and DenseNet-121 for binary classification of brain cancer from 2,931 bright-field transmission optical microscopy images of thin brain tissue. The data come from a commercial TMA, are split at the patient level into training/validation/test cohorts (189/49/74 patients), and are processed with a two-stage transfer-learning protocol, data augmentation, and early stopping. On the held-out test set, DenseNet-121 reaches 88.35% accuracy, 0.9614 precision, 0.8667 recall, and 0.9116 F1-score, while ResNet-50 reaches 82.12%, 0.9035, 0.8142, and 0.8563; the difference is reported as significant by McNemar's test (p = 0.004). The paper concludes that dense connectivity generalizes better on limited medical data.

Significance. If taken at face value, this is a modest but useful empirical benchmark on a relatively small single-site TMA dataset. The patient-level split is an appropriate safeguard against the most obvious data-leakage route, and the reported confusion-matrix numbers are internally consistent: 299 TP + 141 TN = 440 of 498 correct, matching 88.35% accuracy, and the precision, recall, and F1 values are consistent with those counts. The study does not propose a new method or theoretical result, and its value rests entirely on whether the learned signal is genuinely related to cancer morphology rather than slide-preparation or scanner artifacts, and on whether the DenseNet-versus-ResNet gap survives repeated splits and architectural ablations. These conditions are not currently established, so the significance of the headline claim is limited until the robustness analyses are supplied.

major comments (4)

[Section 2.1 and Section 3.1] The manuscript never reports how the 24-core TMA slides are distributed among train, validation, and test sets, nor whether cancerous and non-cancerous cores are balanced within each slide. Because all cores on one slide share staining, sectioning, illumination, and scanner state, a classifier could separate classes by detecting slide-level batch differences rather than tumor morphology. This would inflate both the absolute accuracy and the DenseNet-ResNet gap if the two architectures have different sensitivities to those cues. Please add a slide-level breakdown, per-slide accuracy, and a leave-one-slide-out or slide-stratified cross-validation, or explicitly demonstrate that each slide contains both classes and that per-slide performance is stable. Without this, the 'superior generalization' conclusion in Sections 4 and 5 is not supported.
[Section 2.5 and Section 3.1] The reported 95% confidence intervals are bootstrap intervals over the 498 test predictions, which quantify sampling uncertainty of the test set but not variance due to training initialization, augmentation order, or early stopping. With a single train/validation/test split, the 6.23% accuracy difference (p = 0.004 by McNemar) may not be robust. Please report results across multiple training runs with different random seeds and, ideally, across multiple patient-level splits, giving mean and standard deviation for accuracy, precision, recall, and F1-score. This is needed to establish that the DenseNet advantage is not an artifact of one favorable split or one favorable training run.
[Section 5 and Section 4] The concluding attribution that 'These results underscore the value of dense connectivity for medical image analysis on limited datasets' is a causal claim about architectural design, but only two architectures are compared, and they differ in depth, width, connectivity pattern, and the details of which layers are fine-tuned. The observed gap could stem from any of these differences or from hyperparameter choices. To support the attribution, include an ablation that isolates dense connectivity (for example, a DenseNet variant with reduced connectivity or a controlled set of architectures matched in depth and width), or soften the conclusion to state that DenseNet-121 performed better on this specific dataset without making a general architectural claim.
[Section 2.1 and Section 5] The study uses a single commercial TMA scanned on one microscope, and no external validation cohort is provided. The limitations paragraph acknowledges generalizability concerns, but the abstract and conclusions still state 'superior generalization' and point toward clinical translation. Please either add an external validation set from a different institution or TMA batch, or explicitly restrict the claims to the dataset at hand. As written, the wording overstates what a single-batch study can establish.

minor comments (4)

[Section 2.1] There is a typo: 'TAM samples' should be 'TMA samples'. Also, the citation markers [10,11] and [13,14,16] in this section do not align with the cited works in context; for example, reference [10] is the DenseNet paper and is not about TMA slide design. Please re-check all citation placements.
[Section 3.2 and Figures 2 and 3] The epoch counts are inconsistent: the text says early stopping halted DenseNet-121 at epoch 23 and ResNet-50 at epoch 17, while the caption of Figure 2 says 'over 36 epochs' and the Methods specify Stage 1 for 10 epochs and Stage 2 for 20 epochs. Please reconcile the numbers and make clear which training phase each figure corresponds to.
[Figure 2] The caption describes panel (a) as loss and panel (b) as accuracy, but the panel labels appear reversed relative to the caption. Please verify the panel assignments.
[Section 2.1] The sentence stating that 'the transmission intensity of the microscope is proportional to the mass density of the samples' is repeated with slightly different wording in the same paragraph. Please consolidate and, if this proportionality is load-bearing for the biological interpretation, clarify how it translates into a learnable bright-field signature in the classification setting.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the reported DenseNet-121 vs. ResNet-50 comparison is a measured held-out test result, not a fitted or self-referential derivation.

full rationale

This paper is an empirical benchmarking study, not a derivation chain. Both models are trained on a subject-level split training set, checkpoints are selected by validation loss, and all reported accuracies, precision, recall, F1-scores, and the McNemar p-value are computed on a held-out test set (498 slices). No fitted parameter is renamed as a prediction, and no equation in the paper reduces the test outcome to the training inputs by construction. The only self-citation, reference [16], is used for scanning details and a physical rationale that transmission intensity is proportional to mass density; even if that rationale were contested, the central claim rests on measured classification performance, not on the cited prior work. The comparison between architectures is an actual experiment with standard transfer learning and independent test evaluation, so no circular step is present.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The paper's central comparison rests on standard transfer learning practice, a private dataset with asserted expert labels, and an imaging-intensity-to-density assumption. There are no invented entities or fitted constants in the mathematical sense, but the training hyperparameters are hand-chosen and the domain assumptions (label correctness, imaging signal, no leakage) are not independently verified.

free parameters (6)

Stage 1 learning rate = 1e-4
Chosen by hand for the frozen-feature training phase; affects optimization trajectory and the final comparison.
Stage 2 learning rate = 5e-5
Chosen by hand for fine-tuning; lower rate intended to preserve pretrained features.
Weight decay = 1e-4
Chosen by hand to regularize trainable parameters; standard L2 penalty.
Dropout rate = 0.5
Applied to the classifier head; chosen by hand.
Early stopping patience = 5 epochs
Chosen by hand; determines the stopping point for each model.
Data augmentation ranges = rotation +/-15 degrees, translation +/-5%, intensity jitter +/-10%
Chosen by hand; augmentation strategy influences overfitting and generalization.

assumptions (5)

domain assumption ImageNet-pretrained weights provide useful features for bright-field microscopy images
The transfer learning protocol in Section 2.3 assumes that general natural-image features support the medical imaging task; no analysis verifies which features transfer.
domain assumption Ground-truth labels correctly classify each slice as cancerous or non-cancerous
Section 2.1 states that expert pathologists characterized the TMA samples and neuroradiologists annotated slices, but no inter-observer agreement or label quality metrics are reported.
domain assumption Bright-field transmission intensity is proportional to tissue mass density
Section 2.1 relies on this physical assumption, traced to reference [16], to justify that cancer-related density changes create a learnable signal.
domain assumption The patient-level split prevents data leakage
Section 2.1 reports a patient-level split, but no patient identifiers or verification of non-overlap are provided.
domain assumption The optimization and regularization choices allow a fair architecture comparison
Sections 2.3 and 2.4 specify a single training protocol with hand-chosen hyperparameters; the comparison is only valid under this protocol, and no sensitivity analysis is reported.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Evaluation and optimization of deep learning models for enhanced detection of brain cancer using transmission optical microscopy of thin brain tissue samples." pith.science (2026). https://pith.science/paper/SHOFNS3R

@misc{pith2026250511735,
  author       = {Pith},
  title        = {Pith review of: Evaluation and optimization of deep learning models for enhanced detection of brain cancer using transmission optical microscopy of thin brain tissue samples},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SHOFNS3R}},
  note         = {Machine review of arXiv:2505.11735}
}

read the original abstract

Optical transmission spectroscopy is one method to understand brain tissue structural properties from brain tissue biopsy samples, yet manual interpretation is resource intensive and prone to inter observer variability. Deep convolutional neural networks (CNNs) offer automated feature learning directly from raw brightfield images. Here, we evaluate ResNet50 and DenseNet121 on a curated dataset of 2,931 bright-field transmission optical microscopy images of thin brain tissue, split into 1,996 for training, 437 for validation, and 498 for testing. Our two stage transfer learning protocol involves initial training of a classifier head on frozen pretrained feature extractors, followed by fine tuning of deeper convolutional blocks with extensive data augmentation (rotations, flips, intensity jitter) and early stopping. DenseNet121 achieves 88.35 percent test accuracy, 0.9614 precision, 0.8667 recall, and 0.9116 F1 score the best performance compared to ResNet50 (82.12 percent, 0.9035, 0.8142, 0.8563). Detailed analysis of confusion matrices, training and validation curves, and classwise prediction distributions illustrates robust convergence and minimal bias. These findings demonstrate the superior generalization of dense connectivity on limited medical datasets and outline future directions for multi-class tumor grading and clinical translation.

Figures

Figures reproduced from arXiv: 2505.11735 by the authors.

**Figure 1.** Representative optical transmission microscope images. (a1-a4) Representative control tissue and (b1-b4) represent brain cancer thin tissue brightfield images from the human brains using the TMA tissue samples. It is difficult to see the differences from the bright-field images using eye estimation. Therefore, an AIassisted machine learning approach will be more valuable for cancer monitoring and detection [13,14,1… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 19 canonical work pages

[10]

L.-C. Chen, Y. Zhu, G. Papandreou, F. Schroff, and H. Adam, Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation, in (2018), pp. 801–818

work page 2018
[11]

Squeeze and Excitation

A. G. Roy, N. Navab, and C. Wachinger, Recalibrating Fully Convolutional Networks With Spatial and Channel “Squeeze and Excitation” Blocks, IEEE Transactions on Medical Imaging 38, 540 (2019)

work page 2019
[16]

Sengupta, M

B. Sengupta, M. Alrubayan, M. Kolla, Y. Wang, E. Mallet, A. Torres, R. Solis, H. Wang, and P. Pradhan, AI-Based Detection of Optical Microscopic Images of Pseudomonas aeruginosa in Planktonic and Biofilm States, Information 16, 4 (2025)

work page 2025
[1]

L. A. Torre, F. Bray, R. L. Siegel, J. Ferlay, J. Lortet-Tieulent, and A. Jemal, Global cancer statistics, 2012, CA: A Cancer Journal for Clinicians 65, 87 (2015)

work page 2015
[2]

D. N. Louis, A. Perry, G. Reifenberger, A. von Deimling, D. Figarella-Branger, W. K. Cavenee, H. Ohgaki, O. D. Wiestler, P. Kleihues, and D. W. Ellison, The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary, Acta Neuropathol 131, 803 (2016)

work page 2016
[3]

R. Shi, J. C. Pinto, I. Rienda, P. Caie, C. Eloy, and A. Polónia, Image analysis for bright-field HER2 in situ hybridization: validation for clinical use, Virchows Arch 486, 541 (2025)

work page 2025
[4]

Gulshan et al., Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs, JAMA 316, 2402 (2016)

V. Gulshan et al., Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs, JAMA 316, 2402 (2016)

work page 2016
[5]

Coudray, P

N. Coudray, P. S. Ocampo, T. Sakellaropoulos, N. Narula, M. Snuderl, D. Fenyö, A. L. Moreira, N. Razavian, and A. Tsirigos, Classification and mutation prediction from non–small cell lung cancer histopathology images using deep learning, Nat Med 24, 1559 (2018)

work page 2018

Show all 19 references

[6]

Tajbakhsh, J

N. Tajbakhsh, J. Y. Shin, S. R. Gurudu, R. T. Hurst, C. B. Kendall, M. B. Gotway, and J. Liang, Convolutional Neural Networks for Medical Image Analysis: Full Training or Fine Tuning?, IEEE Transactions on Medical Imaging 35, 1299 (2016)

2016
[7]

Yamashita, M

R. Yamashita, M. Nishio, R. K. G. Do, and K. Togashi, Convolutional neural networks: an overview and application in radiology, Insights Imaging 9, 611 (2018)

2018
[8]

K. He, X. Zhang, S. Ren, and J. Sun, Deep Residual Learning for Image Recognition, in (2016), pp. 770–778

2016
[9]

Huang, Z

G. Huang, Z. Liu, L. van der Maaten, and K. Q. Weinberger, Densely Connected Convolutional Networks, in (2017), pp. 4700–4708

2017
[12]

I. P. G. Y. P. Putra, N. W. J. K. Dewi, P. S. W. Lesmana, I. G. T. Suryawan, and P. S. U. Putra, Comparison of ResNet-50 and DenseNet-121 Architectures in Classifying Diabetic Retinopathy, Indonesian Journal of Data and Science 6, 1 (2025). 10

2025
[13]

AI-Assisted High-Throughput Tissue Microarray Workflow, https://www.mdpi.com/2409- 9279/7/6/96
[14]

Shamai, Y

G. Shamai, Y. Binenbaum, R. Slossberg, I. Duek, Z. Gil, and R. Kimmel, Artificial Intelligence Algorithms to Assess Hormonal Status From Tissue Microarrays in Patients With Breast Cancer, JAMA Network Open 2, e197700 (2019)

2019
[15]

S. A. Harmon et al., High throughput assessment of biomarkers in tissue microarrays using artificial intelligence: PTEN loss as a proof-of-principle in multi-center prostate cancer cohorts, Modern Pathology 34, 478 (2021)

2021
[17]

T. T. L. Vuong, K. Kim, B. Song, and J. T. Kwak, Joint categorical and ordinal learning for cancer grading in pathology images, Medical Image Analysis 73, 102206 (2021)

2021
[18]

Gao et al., An explainable longitudinal multi-modal fusion model for predicting neoadjuvant therapy response in women with breast cancer, Nat Commun 15, 9613 (2024)

Y. Gao et al., An explainable longitudinal multi-modal fusion model for predicting neoadjuvant therapy response in women with breast cancer, Nat Commun 15, 9613 (2024)

2024
[19]

R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, Grad-CAM: Visual Explanations From Deep Networks via Gradient-Based Localization, in (2017), pp. 618–626

2017

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [10]

L.-C. Chen, Y. Zhu, G. Papandreou, F. Schroff, and H. Adam, Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation, in (2018), pp. 801–818

work page 2018

[2] [11]

Squeeze and Excitation

A. G. Roy, N. Navab, and C. Wachinger, Recalibrating Fully Convolutional Networks With Spatial and Channel “Squeeze and Excitation” Blocks, IEEE Transactions on Medical Imaging 38, 540 (2019)

work page 2019

[3] [16]

Sengupta, M

B. Sengupta, M. Alrubayan, M. Kolla, Y. Wang, E. Mallet, A. Torres, R. Solis, H. Wang, and P. Pradhan, AI-Based Detection of Optical Microscopic Images of Pseudomonas aeruginosa in Planktonic and Biofilm States, Information 16, 4 (2025)

work page 2025

[4] [1]

L. A. Torre, F. Bray, R. L. Siegel, J. Ferlay, J. Lortet-Tieulent, and A. Jemal, Global cancer statistics, 2012, CA: A Cancer Journal for Clinicians 65, 87 (2015)

work page 2015

[5] [2]

D. N. Louis, A. Perry, G. Reifenberger, A. von Deimling, D. Figarella-Branger, W. K. Cavenee, H. Ohgaki, O. D. Wiestler, P. Kleihues, and D. W. Ellison, The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary, Acta Neuropathol 131, 803 (2016)

work page 2016

[6] [3]

R. Shi, J. C. Pinto, I. Rienda, P. Caie, C. Eloy, and A. Polónia, Image analysis for bright-field HER2 in situ hybridization: validation for clinical use, Virchows Arch 486, 541 (2025)

work page 2025

[7] [4]

Gulshan et al., Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs, JAMA 316, 2402 (2016)

V. Gulshan et al., Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs, JAMA 316, 2402 (2016)

work page 2016

[8] [5]

Coudray, P

N. Coudray, P. S. Ocampo, T. Sakellaropoulos, N. Narula, M. Snuderl, D. Fenyö, A. L. Moreira, N. Razavian, and A. Tsirigos, Classification and mutation prediction from non–small cell lung cancer histopathology images using deep learning, Nat Med 24, 1559 (2018)

work page 2018