Self-Supervised Radiograph Anatomical Region Classification -- How Clean Is Your Real-World Data?

Daniel Rueckert; Jessica Ritter; Paul Hager; Rickmer Braren; Simon Langer

REVIEW 3 major objections 4 minor 25 references

Self-Supervised Radiograph Anatomical Region Classification -- How Clean Is Your Real-World Data?

T0 review · 3 major / 4 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 This paper shows that contrastive self-supervised pretraining on 48,434 skeletal radiographs classifies 14 anatomical regions at 96.6% accuracy with one model, 97.7% with an ensemble, and uses the model's disagreements to catch wrong PACS…

desk verdict A practically useful label-noise audit undermined by a test-set leakage that makes the headline accuracies unverifiable. read the letter →

arxiv 2412.15967 v1 pith:DNCP55VD submitted 2024-12-20 cs.CV cs.AI

Simon Langer , Jessica Ritter , Rickmer Braren , Daniel Rueckert , Paul Hager This is my paper · ORCID

classification cs.CVcs.AI

keywords self-supervisedlearningcontrastiveradiographclassificationanatomicalregionPACSmetadataqualitylabelnoiseSimCLRmedicalimaging

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 claims that self-supervised contrastive pretraining on 48,434 skeletal radiographs is enough to classify the anatomical region into 14 classes with 96.6% accuracy using a single model and 97.7% using an ensemble. It also claims that only 310 labeled images (1% of the training set) suffice to reach 92.2% accuracy, making the approach usable in low-resource settings. The central application is data cleaning: an expert radiologist's review of the model's 328 test errors found that 35% of them were wrong PACS labels and 11% were out-of-domain images, and 98 of the 116 wrongly labeled images were actually predicted correctly by the model. Accounting for these label errors raises the corrected accuracy to 98.0% for the single model and 98.8% for the ensemble. If true, this means existing hospital metadata can be audited and corrected from the model's disagreements.

What carries the argument

The load-bearing mechanism is contrastive self-supervised pretraining: SimCLR trains a ResNet18 backbone to pull together two augmentations of the same radiograph and push apart other images in a batch, while BYOL removes the negative pairs and trains an online network to predict a target network's representation. A supervised contrastive variant extends positive pairs to all images sharing the same PACS label. After pretraining, the frozen backbone is evaluated with a single linear layer, and an ensemble averages the softmax outputs of the three pretrained models. Two custom preprocessing steps support the claim: border removal and rotation normalization reduce reliance on image-frame artifacts, and a novel augmentation that randomly inserts zero to two surgery-planning gauges at random locations and scales deliberately prevents the model from using gauges as a shortcut, which the paper verifies with gradient-based heatmaps.

What would settle it

Randomly sample a few hundred test-set images in which the model and the PACS label agree, have an expert radiologist independently re-label them, and recompute accuracy on the re-labeled set; if a substantial share of these agreements are actually mislabeled by both PACS and the model, the theoretical 98.0% accuracy does not hold, whereas an error rate near zero would confirm it.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that contrastive self-supervised pretraining on unlabeled skeletal radiographs produces anatomical region features so strong that a single linear layer on top reaches 96.6% test accuracy across 14 classes, and averaging the softmax outputs of three such models (SimCLR, BYOL, and supervised contrastive) reaches 97.7%. The same features are label-efficient: training the linear layer on just 1% of the training labels (310 images) yields 92.2%, far above the 57.1% of a fully supervised baseline trained from scratch. The paper further establishes that the remaining errors are largely label noise rather than model failures: an expert radiologist re-labeling the 328 disagreement cases found 116 incorrect PACS labels, 36 out-of-domain images, and 2 unusable images, with 98 of the 116 incorrect labels being images the model had classified correctly. After filtering and correcting these labels, the theoretical accuracy becomes 98.0% for the SimCLR model and 98.8% for the ensemble. The conclusion the authors draw is that a self-supervised model can be used as a practical tool to find and fix data-entry errors in PACS metadata.

Load-bearing premise

The corrected accuracy of 98.0% assumes that the 328 images where the model and the PACS label disagreed are the only test-set errors, so fixing those labels fully accounts for label noise; the roughly 9,400 test images where the model and label agreed were never audited.

Editorial extensions

If this is right

A hospital with only a few hundred labeled radiographs can train a usable anatomical region classifier, since 1% of the labels (310 images) reach 92.2% accuracy.
PACS metadata can be audited by comparing model predictions to stored labels; the spine, shoulder, and clavicle classes are the most inconsistent and would benefit first.
External datasets that arrive without anatomical metadata can be automatically labeled, enabling their inclusion in research cohorts.
Downstream models that are only valid for specific body regions can reject or reroute out-of-domain inputs using the predicted region, reducing silent misdiagnosis.
The ensemble's 97.7% accuracy on raw PACS labels implies that most residual 'errors' in real-world radiograph metadata are label corruption, not model failure.

Reading between the lines

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

The reported corrected accuracy of 98.0% is conditional on the audit being restricted to the 328 cases where the model disagreed with the PACS label; the roughly 9,400 test images where they agreed were never re-examined, so the true accuracy could be lower if some of those labels are wrong in the same way as the model.
The same disagreement-audit procedure could be applied to the training and validation splits, effectively cleaning the entire dataset; the paper notes its architecture allows this but does not carry it out.
Because the gauge augmentation was designed to combat a dataset-specific shortcut, the method's transfer to other radiograph collections may depend on whether similar domain artifacts are handled in the same way.
The claim that the model corrects data-entry mistakes is about agreement between model and label, not about the model being a perfect oracle; a future test could compare the model's corrected labels against a second independent re-labeling to measure agreement.

Share X Bluesky LinkedIn Reddit HN

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

The paper is worth a look for one reason: the expert radiologist audit of model failures is a genuinely useful way to expose noisy PACS metadata, and the gauge augmentation is a thoughtful, targeted fix for a real shortcut. Those contributions stand on their own. The problem is that the headline numbers—96.6% single-model, 97.7% ensemble, and the 92.2% 1% label result—are not trustworthy because the self-supervised backbone appears to be pretrained on the full 48,434-image dataset, including the test split. Figure 1 makes this explicit: “Self-Supervised Pre-training (all data).” Section 3.1 defines the train/validation/test split, but Section 3.3 describes pretraining on all radiographs without restricting to the training split. That is transductive learning: the feature extractor has already seen the test images before the linear probe is trained. It inflates linear evaluation accuracy and makes the comparison against the fully supervised baseline, which only trains on labeled subsets, unfair. The low-label comparison inherits the same problem. The corrected accuracy of 98.0% is also selection-biased: the expert only audited the 328 images where the model disagreed with the PACS label, not the thousands where both agreed. Both-wrong agreements would lower the corrected number. I don't think these are deliberate deceptions—the paper is transparent enough about the all-data pretraining—but they are load-bearing flaws. The method probably works, and the label-noise detection idea is solid, but the empirical claims as reported are not yet supported. There are also minor issues: no error bars, single-center data, no code or data release, and the low-label experiment lacks repeated-seed variance. The paper deserves a serious referee because the question—how clean is real-world PACS data, and can self-supervised models catch the dirt—is important for operational medical AI, and the audit procedure is a legitimate contribution. But the evaluation protocol has to be corrected: retrain the backbone on the training split only, re-report all numbers, and ideally validate on an external dataset. If the numbers hold up after that, this becomes a useful practical paper. As is, I would not cite the headline accuracy; I might cite the audit methodology.

Referee Report

3 major / 4 minor

Summary. The paper addresses automatic classification of skeletal radiographs into 14 anatomical regions from a real-world PACS dataset of 48,434 images. The authors pretrain ResNet18 backbones with SimCLR, BYOL, and supervised contrastive learning, then evaluate frozen-backbone linear probes. They report a single-model test accuracy of 96.6%, an ensemble accuracy of 97.7%, and 92.2% accuracy using only 1% of the training labels. They also report an expert-radiologist audit of the single model's 328 test errors, finding that 35% of these were incorrect PACS labels and 11% were out-of-domain images, which leads to a claimed 'theoretical' accuracy of 98.0% (single) and 98.8% (ensemble) after relabeling. The paper further proposes using the model to detect and correct noisy PACS metadata.

Significance. If the reported numbers hold under a clean evaluation protocol, the work would be practically valuable: it demonstrates that self-supervised representation learning can enable anatomical-region classification with very few labels and that learned models can flag data-entry errors in hospital PACS systems. The authors deserve credit for using a large in-house clinical dataset, for explicitly targeting label noise as a research question, and for obtaining independent expert-radiologist labels to ground the noise analysis. However, the central quantitative claims are weakened by a potentially serious evaluation-protocol issue: the self-supervised pretraining appears to include the test set as unlabeled data, and the supervised contrastive model appears to use test labels during pretraining. This affects the headline accuracies, the low-label results, and the error audit, and it must be resolved before the results can be accepted as estimates of held-out generalization.

major comments (3)

[Section 3.3 and Figure 1] The pretraining protocol appears to leak test-set information. Section 3.1 defines a 64/16/20 train/validation/test split of the 48,434 radiographs, but Section 3.3 states that pretraining is performed on the full 48,434 radiographs without excluding the test split, and Figure 1 explicitly labels 'Self-Supervised Pre-training (all data)'. Moreover, the supervised contrastive method is described as using label information 'for the whole dataset', meaning test labels are used during pretraining for the SupCon model that contributes to the ensemble. Under this protocol, the reported 96.6% single-model and 97.7% ensemble accuracies are transductive estimates, not held-out generalization, and the comparison against the fully supervised baseline trained only on labeled training subsets is unfair. The authors must rerun the pretraining using only the 31,011-image training split (and the validation split if needed, with the test split strictly held out) and report whether the main conclusions change.
[Section 4.4 and Table 2] The corrected 'theoretical' accuracy is not established because the expert audit is conditioned on the model's disagreements. The radiologist audited only the 328 test images where the SimCLR prediction differed from the PACS label, leaving the roughly 9,400 test images where model and PACS label agreed unaudited. Errors in which both the model and the PACS label are wrong would not appear in the 328 cases, so the claim that correcting labels raises accuracy to 98.0% (and 98.8% for the ensemble) is an upper bound, not a measured value. To support the data-cleaning claim, the authors should audit a random sample of the agreement cases (or an independent random sample of the full test set) and report the actual error rate after relabeling. This is a load-bearing issue for the paper's stated application of correcting PACS metadata.
[Section 4.1 and Figure 2] The low-label experiment inherits the same test-set leakage: the SimCLR and BYOL backbones were pretrained on the test images, so the 92.2% accuracy at 1% of training labels reflects features that have already seen the test distribution. Additionally, the supervised baseline is described only as a 'randomly initialized ResNet18 trained fully supervised'; the paper does not specify whether this baseline uses the same image cleaning, gauge augmentation, and SimCLR-style augmentations, or the same optimization schedule. Without these details and without a leakage-free pretraining, the large gap between the self-supervised and supervised curves cannot be attributed to self-supervision. Please repeat the low-label sweep with a backbone pretrained only on the training split and report the baseline setup precisely.

minor comments (4)

[Section 3.3 (BYOL)] The sentence 'accordingly, we set τ_base = 0.9995 as recommended in [8]' refers to BYOL's exponential moving average decay, not to the temperature τ used in SimCLR; using the same symbol for both quantities is confusing. Please rename the BYOL EMA parameter (e.g., m or target_ema) and clarify the values.
[Section 4.4 and Figure 4] The confusion matrices in Figure 4 are difficult to read: diagonal entries are omitted in panel (a), and the delta matrix in panel (b) uses colors without a numeric legend for the magnitude of changes. Please add clear numeric annotations and a caption explaining how to read the 'delta' matrix.
[Supplementary Table 1] The row 'Gauge Occurrences 1/3 probability for 0, 1, 2' is ambiguous: it should state explicitly that with probability 1/3 each, zero, one, or two gauges are inserted, and whether the same distribution is used in pretraining and training.
[Section 3.2] The sentence 'we add a novel augmentation where we use 6 example gauges extracted from the dataset' should clarify whether those 6 example gauges are extracted from the training split only or from the entire dataset, since using test images to build augmentations would be another form of leakage.

Circularity Check

2 steps flagged · score 6.0 of 10

Two evaluation steps are partially circular: pretraining appears to use all data including test images, and the corrected accuracy is derived from model-selected disagreement cases.

fitted input called prediction [Fig. 1 overview; Sec. 3.1 dataset split; Sec. 3.3 pretraining]
"Fig.1: Overview – we pre-train our backbone using self-supervision, then train a fully connected head, and finally use its high quality predictions to correct noisy PACS labels. ... Self-Supervised Pre-training (all data). ... A total of 48,434 radiographs with corresponding anatomical region labels from our hospital PACS were exported as DICOM files. ... Subsequently, we split the dataset into 31,011 (64%)/7,677 (16%)/9,746 (20%) training/validation/test images."

If the figure is literal, the self-supervised backbone is pretrained on all 48,434 radiographs, which includes the 9,746 test images, before the linear head is trained and evaluated on that same test split. The reported 96.6% single-model and 97.7% ensemble accuracies (and the 92.2% low-label result) are therefore in-sample for the feature extractor: the test images are an input to the pretraining, so the test 'prediction' is not an estimate of generalization to unseen radiographs. No section explicitly restricts pretraining to the 31,011-image training split.
other [Sec. 4.4]
"we tasked an expert radiologist with re-evaluating all test failure cases of the SimCLR model (n=328), almost half of which turned out to be either incorrectly labeled (n=116), out-of-domain (i.e. not in the 14 classes) (n=36), or of unusable quality (n=2)... A total of 98 of the 116 incorrectly labeled in-domain images are in fact correctly predicted by our SimCLR model, increasing its theoretical accuracy (when excluding out-of-domain and unusable images) to 98.0%"

The corrected 'theoretical accuracy' is computed by re-labeling only the 328 images where the model and PACS labels disagree and then counting the model's predictions as correct. The ~9,400 test images where the model and PACS agree are never audited. Because the model was trained on the same noisy PACS labels, the disagreement set is selected by the model itself; any PACS label error the model reproduces is invisible. Thus the 98.0%/98.8% figures are not estimates of true accuracy on the test set; they are conditional on the model's own error set.

full rationale

The paper's training objectives are not circular in themselves: SimCLR, BYOL, and supervised contrastive losses do not define the target classes in terms of the reported accuracies, and the linear probe uses genuinely independent PACS labels. There are no load-bearing self-citations: references to SAM-X and DeepMCAT by Rueckert are contextual, not used to justify the reported numbers. The circularity is in the evaluation protocol rather than in the learning rule. Step 1 rests on Figure 1's explicit 'all data' and the absence of any restriction of pretraining to the training split; if the authors actually pretrained only on the training split, this step would be moot, but as written the test set is presented as an input to the backbone. Step 2 rests on the audit being restricted to the model's disagreement set; the radiologist labels are external and valuable, but the recomputed accuracy assumes no errors among the thousands of agreement cases. These two issues make the central quantitative claims partially in-sample and partially self-referential, warranting a 6 rather than a higher score because the radiologist audit and the contrastive training itself retain independent content.

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

The central claim does not introduce new physical entities. It rests on standard SSL machinery, a domain-specific augmentation, and a noisy-label evaluation whose main unstated burden is the assumption that the audited failure subset is sufficient to estimate true accuracy.

free parameters (5)

Gauge insertion probability = uniform over 0, 1, or 2 gauges per image
Hand-chosen; no ablation validates this specific distribution.
Gauge scale and opacity ranges = scale [0.8, 1.2], opacity [0.75, 1]
Hand-chosen to mimic real planning gauges; no sensitivity analysis.
Pretraining learning rate = 3e-4
Standard Adam setting from prior SSL work, not tuned for this task.
Pretraining epochs = 1000
Long schedule chosen for convergence; no early-stopping analysis.
Number of example gauges = 6
Six gauges extracted from the dataset; not clear they cover the test-set gauge variability.

assumptions (5)

domain assumption PACS anatomical region labels are reliable enough to serve as training targets and as a baseline evaluation benchmark.
The model is trained and initially evaluated on these labels; the audit in Section 4.4 later shows a substantial error rate, so this axiom is partially violated.
domain assumption The 14-class anatomical region taxonomy is appropriate and the train/validation/test split is representative of real PACS data.
The task is defined by this taxonomy, and out-of-domain images are excluded post hoc (Section 4.4).
domain assumption The expert radiologist's re-labels of 328 failure cases are the ground truth.
The corrected accuracy in Table 2 relies entirely on these labels; no inter-observer metric is reported.
standard math The standard SSL losses (SimCLR InfoNCE, BYOL MSE) and the linear evaluation protocol measure representation quality.
Taken from prior literature [4,8] without modification.
ad hoc to paper Inserting synthetic gauges at random locations teaches the model to ignore real gauges without harming other features.
Supported only by a qualitative Grad-CAM comparison (Fig. 3); the quantitative accuracy gain is small and only for the pelvis/hip class.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Self-Supervised Radiograph Anatomical Region Classification -- How Clean Is Your Real-World Data?." pith.science (2026). https://pith.science/paper/DNCP55VD

@misc{pith2026241215967,
  author       = {Pith},
  title        = {Pith review of: Self-Supervised Radiograph Anatomical Region Classification -- How Clean Is Your Real-World Data?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DNCP55VD}},
  note         = {Machine review of arXiv:2412.15967}
}

read the original abstract

Modern deep learning-based clinical imaging workflows rely on accurate labels of the examined anatomical region. Knowing the anatomical region is required to select applicable downstream models and to effectively generate cohorts of high quality data for future medical and machine learning research efforts. However, this information may not be available in externally sourced data or generally contain data entry errors. To address this problem, we show the effectiveness of self-supervised methods such as SimCLR and BYOL as well as supervised contrastive deep learning methods in assigning one of 14 anatomical region classes in our in-house dataset of 48,434 skeletal radiographs. We achieve a strong linear evaluation accuracy of 96.6% with a single model and 97.7% using an ensemble approach. Furthermore, only a few labeled instances (1% of the training set) suffice to achieve an accuracy of 92.2%, enabling usage in low-label and thus low-resource scenarios. Our model can be used to correct data entry mistakes: a follow-up analysis of the test set errors of our best-performing single model by an expert radiologist identified 35% incorrect labels and 11% out-of-domain images. When accounted for, the radiograph anatomical region labelling performance increased -- without and with an ensemble, respectively -- to a theoretical accuracy of 98.0% and 98.8%.

Figures

Figures reproduced from arXiv: 2412.15967 by the authors.

**Figure 2.** Impact of the amount of labeled data on final performance. This shows [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Guided GradCam [23,17] visualization of our SimCLR model trained with [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Confusion matrices (CM) describing the results of our quality assessment [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

**Figure 1.** Figure 1: Evolution of an image – the strengths of color jitter, random affine [PITH_FULL_IMAGE:figures/full_fig_p011_1.png]

**Figure 2.** Figure 2: T-SNE visualization of the test images’ features, generated by our Sim [PITH_FULL_IMAGE:figures/full_fig_p011_2.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 17 canonical work pages

[1]

2021 IEEE/CVF International Conference on Computer Vision (ICCV) pp

Azizi, S., Mustafa, B., Ryan, F., Beaver, Z., von Freyberg, J., Deaton, J., Loh, A., Karthikesalingam, A., Kornblith, S., Chen, T., Natarajan, V., Norouzi, M.: Big self-supervised models advance medical image classification. 2021 IEEE/CVF International Conference on Computer Vision (ICCV) pp. 3458–3468 (2021)

work page 2021
[2]

Bradski, G.: The OpenCV Library. Dr. Dobb’s Journal of Software Tools (2000)

2000
[3]

ArXivabs/1807.05520 (2018)

Caron, M., Bojanowski, P., Joulin, A., Douze, M.: Deep clustering for unsupervised learning of visual features. ArXivabs/1807.05520 (2018)

arXiv 2018
[4]

ArXivabs/2002.05709 (2020)

Chen, T., Kornblith, S., Norouzi, M., Hinton, G.E.: A simple framework for con- trastive learning of visual representations. ArXivabs/2002.05709 (2020)

arXiv 2020
[5]

European Radiology31, 1812 – 1818 (2020)

Dratsch, T., Korenkov, M., Zopfs, D., Brodehl, S., Baessler, B., Giese, D., Brinkmann, S., Maintz, D., dos Santos, D.P.: Practical applications of deep learn- ing: classifying the most common categories of plain radiographs in a pacs using a neural network. European Radiology31, 1812 – 1818 (2020)

work page 2020
[6]

Journal of Digital Imaging34, 66 – 74 (2020)

Fang, X., Harris, L., Zhou, W., Huo, D.: Generalized radiographic view identifica- tion with deep learning. Journal of Digital Imaging34, 66 – 74 (2020)

work page 2020
[7]

German Federal Office for Radiation Protection (BFS): X-ray diagnos- tics: Frequency and radiation exposure of the german population (2022), https://www.bfs.de/EN/topics/ion/medicine/diagnostics/x-rays/ frequency-exposure.html, accessed: 2023-06-12

work page 2022
[8]

ArXiv abs/2006.07733 (2020)

Grill, J.B., Strub, F., Altch’e, F., Tallec, C., Richemond, P.H., Buchatskaya, E., Doersch,C.,Pires,B.Á.,Guo,Z.D.,Azar,M.G.,Piot,B.,Kavukcuoglu,K.,Munos, R., Valko, M.: Bootstrap your own latent: A new approach to self-supervised learn- ing. ArXiv abs/2006.07733 (2020)

arXiv 2020

Show all 25 references

[9]

In: SPIE Medical Imaging (2002)

Gueld, M.O., Kohnen, M., Keysers, D., Schubert, H., Wein, B.B., Bredno, J., Lehmann, T.M.: Quality of dicom header information for image categorization. In: SPIE Medical Imaging (2002)

2002
[10]

2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) pp

He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) pp. 770–778 (2015)

2015
[11]

European Radiology33, 1537 – 1544 (2022)

Hinterwimmer, F., Consalvo, S., Wilhelm, N.J., Seidl, F., Burgkart, R., von Eisenhart-Rothe, R., Rueckert, D., Neumann, J.: SAM-X: sorting algorithm for musculoskeletal x-ray radiography. European Radiology33, 1537 – 1544 (2022)

2022
[12]

In: International Conference on Machine Learning (2015)

Ioffe, S., Szegedy, C.: Batch normalization: Accelerating deep network training by reducing internal covariate shift. In: International Conference on Machine Learning (2015)

2015
[13]

European Radiology32, 8769 – 8776 (2022)

Jonske, F., Dederichs, M., Kim, M.S., Keyl, J., Egger, J., Umutlu, L., Forsting, M., Nensa, F., Kleesiek, J.: Deep learning–driven classification of external dicom studies for pacs archiving. European Radiology32, 8769 – 8776 (2022)

2022
[14]

In: Engelhardt, S., Oksuz, I., Zhu, D., Yuan, Y., Mukhopadhyay, A., Heller, N., Huang, S.X., Nguyen, H., Sznitman, R., Xue, Y

Kart, T., Bai, W., Glocker, B., Rueckert, D.: DeepMCAT: Large-scale deep clus- tering for medical image categorization. In: Engelhardt, S., Oksuz, I., Zhu, D., Yuan, Y., Mukhopadhyay, A., Heller, N., Huang, S.X., Nguyen, H., Sznitman, R., Xue, Y. (eds.) Deep Generative Models,...

2021
[15]

In: Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., Lin, H

Khosla, P., Teterwak, P., Wang, C., Sarna, A., Tian, Y., Isola, P., Maschinot, A., Liu, C., Krishnan, D.: Supervised contrastive learning. In: Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., Lin, H. (eds.) Advances in Neural In- formation Processing Systems. vol. 33, pp....

2020
[16]

In: Bengio, Y., LeCun, Y

Kingma, D.P., Ba, J.: Adam: A method for stochastic optimization. In: Bengio, Y., LeCun, Y. (eds.) 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings (2015), http://arxiv.org/abs/1412.6980

2015 arXiv
[17]

ArXiv abs/2009.07896 (2020)

Kokhlikyan, N., Miglani, V., Martin, M., Wang, E., Alsallakh, B., Reynolds, J., Melnikov, A., Kliushkina, N., Araya, C., Yan, S., Reblitz-Richardson, O.: Cap- tum: A unified and generic model interpretability library for pytorch. ArXiv abs/2009.07896 (2020)

2020 arXiv
[18]

Radiology p

Larson, D.B., Magnus, D.C., Lungren, M.P., Shah, N.H., Langlotz, C.: Ethics of using and sharing clinical imaging data for artificial intelligence: A proposed frame- work. Radiology p. 192536 (2020)

2020
[19]

arXiv: Learning (2016)

Loshchilov, I., Hutter, F.: Sgdr: Stochastic gradient descent with warm restarts. arXiv: Learning (2016)

2016
[20]

Journal of Machine Learning Research 9, 2579–2605 (2008)

van der Maaten, L., Hinton, G.E.: Visualizing data using t-sne. Journal of Machine Learning Research 9, 2579–2605 (2008)

2008
[21]

In: 5th Berkeley Symp

MacQueen, J.: Some methods for classification and analysis of multivariate obser- vations. In: 5th Berkeley Symp. Math. Statist. Probability. pp. 281–297. University of California Los Angeles (1967)

1967
[22]

ArXivabs/1912.01703 (2019), https://arxiv.org/pdf/1912.01703.pdf, https://pytorch.org/, Accessed: 2023-06-12

Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., Köpf, A., Yang, E., De- Vito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., Bai, J., Chintala, S.: PyTorch: An imperat...

2019 arXiv
[23]

In- ternational Journal of Computer Vision128, 336–359 (2016)

Selvaraju, R.R., Das, A., Vedantam, R., Cogswell, M., Parikh, D., Batra, D.: Grad- cam: Visual explanations from deep networks via gradient-based localization. In- ternational Journal of Computer Vision128, 336–359 (2016)

2016
[24]

The American journal of tropical medicine and hygiene88 4, 608–13 (2013)

Tinto, H., Noor, R.A., Wanga, C.L., Valéa, I., Mbaye, M.N., d’Alessandro, U., Ravinetto,R.:Goodclinicalpracticeinresource-limitedsettings:translatingtheory into practice. The American journal of tropical medicine and hygiene88 4, 608–13 (2013)

2013
[25]

Willett, J.: Imaging in trauma in limited-resource settings: A literature review. African Journal of Emergency Medicine9, S21 – S27 (2018) Self-Supervised Radiograph Anatomical Region Classification – How Clean Is Your Real-World Data? (Supplementary Material) (a) Raw Image (b...

2018

Pith tools

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

[1] [1]

2021 IEEE/CVF International Conference on Computer Vision (ICCV) pp

Azizi, S., Mustafa, B., Ryan, F., Beaver, Z., von Freyberg, J., Deaton, J., Loh, A., Karthikesalingam, A., Kornblith, S., Chen, T., Natarajan, V., Norouzi, M.: Big self-supervised models advance medical image classification. 2021 IEEE/CVF International Conference on Computer Vision (ICCV) pp. 3458–3468 (2021)

work page 2021

[2] [2]

Bradski, G.: The OpenCV Library. Dr. Dobb’s Journal of Software Tools (2000)

2000

[3] [3]

ArXivabs/1807.05520 (2018)

Caron, M., Bojanowski, P., Joulin, A., Douze, M.: Deep clustering for unsupervised learning of visual features. ArXivabs/1807.05520 (2018)

arXiv 2018

[4] [4]

ArXivabs/2002.05709 (2020)

Chen, T., Kornblith, S., Norouzi, M., Hinton, G.E.: A simple framework for con- trastive learning of visual representations. ArXivabs/2002.05709 (2020)

arXiv 2020

[5] [5]

European Radiology31, 1812 – 1818 (2020)

Dratsch, T., Korenkov, M., Zopfs, D., Brodehl, S., Baessler, B., Giese, D., Brinkmann, S., Maintz, D., dos Santos, D.P.: Practical applications of deep learn- ing: classifying the most common categories of plain radiographs in a pacs using a neural network. European Radiology31, 1812 – 1818 (2020)

work page 2020

[6] [6]

Journal of Digital Imaging34, 66 – 74 (2020)

Fang, X., Harris, L., Zhou, W., Huo, D.: Generalized radiographic view identifica- tion with deep learning. Journal of Digital Imaging34, 66 – 74 (2020)

work page 2020

[7] [7]

German Federal Office for Radiation Protection (BFS): X-ray diagnos- tics: Frequency and radiation exposure of the german population (2022), https://www.bfs.de/EN/topics/ion/medicine/diagnostics/x-rays/ frequency-exposure.html, accessed: 2023-06-12

work page 2022

[8] [8]

ArXiv abs/2006.07733 (2020)

Grill, J.B., Strub, F., Altch’e, F., Tallec, C., Richemond, P.H., Buchatskaya, E., Doersch,C.,Pires,B.Á.,Guo,Z.D.,Azar,M.G.,Piot,B.,Kavukcuoglu,K.,Munos, R., Valko, M.: Bootstrap your own latent: A new approach to self-supervised learn- ing. ArXiv abs/2006.07733 (2020)

arXiv 2020