Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Medical Slice Transformer: Improved Diagnosis and Explainability on 3D Medical Images with DINOv2

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A slice transformer adapts 2D DINOv2 to 3D scans and beats 3D ResNet

desk verdict Useful architecture paper whose headline accuracy claim is compromised by a likely patient-level data leak in two of three datasets. read the letter →

arxiv 2411.15802 v1 pith:IPNYRJN3 submitted 2024-11-24 eess.IV cs.AIcs.CV

Gustav Müller-Franzes , Firas Khader , Robert Siepmann , Tianyu Han , Jakob Nikolas Kather , Sven Nebelung , Daniel Truhn This is my paper · ORCID
classification eess.IVcs.AIcs.CV
keywords MedicalSliceTransformerDINOv23Dimageanalysisself-supervisedlearningexplainabilitysaliencymapsMRICT
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 the Medical Slice Transformer (MST), which applies the 2D self-supervised model DINOv2 to each slice of a 3D volume and then aggregates the slices with a Transformer, outperforms a standard 3D ResNet on three diagnostic tasks: breast MRI, chest CT, and knee MRI. The reported AUC improvements are 0.94 vs 0.91 for breast, 0.95 vs 0.92 for chest, and 0.85 vs 0.69 for knee, with the breast and knee differences reaching statistical significance. The same attention mechanism that drives classification also produces saliency maps: multiplying the slice-level attention of the Transformer with the within-slice attention of DINOv2 highlights both the relevant slice and the core of the lesion. A blinded radiologist rated these MST maps as correctly locating the lesion slice in 136 of 150 cases and the lesion core in 57 of 150 cases, against 37 and 0 for ResNet's Grad-CAM maps. If confirmed, the approach matters because it lets 2D self-supervised models that need no 3D annotations carry the diagnostic and localization burden in volumetric imaging.

What carries the argument

The central object is the Medical Slice Transformer (MST), a two-stage architecture in which a pre-trained 2D image encoder (DINOv2) converts each slice of a 3D volume into a feature vector, and a sequence Transformer aggregates those vectors with a classification token. The mechanism carrying the argument is the combined attention map: the Transformer's normalized attention over slices tells which slices contain the finding, the DINOv2 encoder's patch attention tells where within the slice, and their product produces a saliency map aligned to the input volume. This single mechanism simultaneously enables classification and explainability, which is what lets the paper claim that MST addresses both the annotation-scarcity and black-box problems in 3D medical imaging.

What would settle it

Have a second panel of radiologists rate the identical saliency maps with model identity hidden; if the 136 of 150 versus 37 of 150 slice-correctness split does not reproduce, the explainability claim is falsified. A quantitative check would compute overlap between MST combined attention maps and expert lesion segmentations on the same volumes.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that the MST framework — a DINOv2 encoder that maps each 2D slice to a feature vector plus a single-layer Transformer that reads the sequence of slice vectors — transfers DINOv2's strong 2D representations to 3D volumes and makes the model's reasoning visible. The classification token of the slice Transformer yields the diagnosis, and its attention over slices, multiplied by DINOv2's patch-level attention, yields a combined saliency map that localizes findings. The paper reports that MST beats 3D ResNet50 in AUC on all three datasets and that its saliency maps are markedly more precise, with Grad-CAM rated as pointing at lesion cores in 0 of 150 cases versus 57 of 150 for MST. Ablations show the Transformer aggregation, not the DINOv2 features alone, is what carries the gain: replacing the Transformer with a linear layer or averaging feature vectors lowers AUC, and freezing DINOv2's weights collapses performance on breast and chest.

Load-bearing premise

The explainability claim rests on one radiologist's blinded ratings of 150 saliency maps; if other radiologists would not reproduce those ratings, the claim of improved explainability loses its support.

Editorial extensions

If this is right

  • 2D self-supervised models pretrained on natural images can be lifted to 3D medical diagnosis without 3D pretraining, reducing the need for large annotated volumetric datasets.
  • A single forward pass yields both a diagnosis and a coarse lesion localization, so the explainability signal does not require a separate segmentation or detection model.
  • The slice Transformer is the load-bearing component: replacing it with linear aggregation or feature averaging lowers AUC, and freezing the DINOv2 backbone degrades performance sharply on breast MRI and chest CT.
  • On the knee MRI task the gain is large (0.85 vs 0.69 AUC), suggesting the benefit is greatest where the 3D context across slices carries diagnostic information.
  • Attention-based saliency maps point to lesion cores in 57 of 150 cases where Grad-CAM points to none, which if reproducible would give clinicians a localization cue that convolution-based heatmaps do not provide.

Reading between the lines

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

  • I read the chest CT result (P=0.13) as the paper's weak statistical link; the burden of proof for the diagnostic claim rests mainly on breast and knee, and a replication with more chest cases would sharpen the picture.
  • A natural extension the paper does not test is using the combined attention map as a weakly supervised segmentation, e.g., comparing it to expert outlines with an overlap metric; the paper only measures slice- and core-level correctness.
  • Because the explainability rating involved a single radiologist, an obvious replication is an inter-observer study with several radiologists; if ratings vary widely, the explainability advantage would need to be redefined.
  • The ablation showing frozen DINOv2 collapses to 0.62 AUC on breast suggests that fine-tuning the 2D encoder to medical data does much of the work; a separate experiment varying the amount of fine-tuning data would separate the contribution of pretraining from the contribution of the Transformer aggregation.
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 / 4 minor

Summary. The paper introduces the Medical Slice Transformer (MST), a framework that adapts the 2D self-supervised foundation model DINOv2 to 3D medical image classification by processing individual slices with a 2D encoder and aggregating slice features with a lightweight Transformer. The authors evaluate MST against a 3D ResNet50 on three public datasets (breast MRI, chest CT, knee MRI) for diagnosis, reporting higher AUC for MST on all three, with P-values of 0.02, 0.13, and 0.001 respectively. They also present a radiologist's subjective evaluation of saliency maps, reporting that MST attention maps are more often correct in highlighting the slice and lesion core than ResNet Grad-CAM maps. The paper includes ablation studies on the slice aggregation mechanism, positional encoding, backbone size, freezing, and register tokens, and provides public code and data links.

Significance. If the central claims hold, the paper would be a useful contribution: it demonstrates a surprisingly simple and parameter-efficient way to repurpose 2D self-supervised vision transformers for 3D medical imaging, with favorable diagnostic performance and potentially more interpretable attention maps than post-hoc Grad-CAM. The strengths include the use of multiple public datasets, a controlled preprocessing pipeline, head-to-head comparison with a standard 3D CNN, and a thorough ablation study. The code is publicly available, which supports reproducibility. However, the significance is tempered by two load-bearing issues: the chest CT comparison is not statistically significant, and the train/test split for the breast and chest datasets may not be patient-stratified, which could invalidate the significance tests. The explainability claim rests on a single-reader subjective assessment without inter-rater reliability.

major comments (3)
  1. [Methods, Model Implementation and Training; Methods, Dataset Collection and Preprocessing] The train/test split is described as an 80/20 split stratified by label distribution, but the manuscript does not state that the split was performed at the patient level. For DUKE, each MRI study is separated into left and right breast stacks, and for LIDC-IDRI the classification unit is the individual nodule (1,625 nodules from 722 patients). If the split is applied to stacks or nodules rather than patients, the same patient can contribute samples to both training and test sets. This creates data leakage that can inflate AUC and violates the independence assumption underlying DeLong's test, potentially making the reported P=0.02 for breast unreliable. Please clarify the splitting unit; if it is not patient-level, re-run the experiments with patient-stratified splitting and report whether the breast and chest results still hold.
  2. [Results, MST Outperforms Standard CNN Architectures; Abstract; Discussion] The chest CT comparison yields P=0.13, which is not statistically significant at the conventional 0.05 threshold. The abstract and conclusion state that "MST achieved higher AUC values compared to ResNet across all three datasets," which overstates the evidence: only the knee result is significant as reported, and the breast result is significant only if the patient-leakage issue is resolved. Please qualify the claim by reporting the chest result as a non-significant trend, and consider presenting effect sizes and confidence intervals for all three comparisons.
  3. [Results, MST Provides Better Model Explainability; Methods, Saliency Maps] The explainability evaluation is based on a single radiologist's blinded review of 50 saliency maps per dataset, with no inter-rater reliability measure (e.g., Cohen's kappa) and no quantitative localization metric (e.g., overlap with lesion masks). Because attention maps and Grad-CAM images are visually distinct, the rater may be effectively unblinded. This evidence is insufficient to support the claim of "substantial improvement in model explainability." Please add multiple readers with agreement statistics or a quantitative evaluation against ground-truth lesion locations, or temper the conclusion to reflect the single-reader nature of the assessment.
minor comments (4)
  1. [Tables 2 and 3] The table captions state "Results are shown as mean accuracy ± standard deviation," but the values reported (e.g., 0.94±0.01 for breast MRI) are evidently AUC values, matching the AUC results in the text and Figure 2. Please correct the captions to say "AUC" or clarify the metric.
  2. [Methods, Saliency Maps] The sentence "The color coding toward blue indicates low attention, while the spectrum toward red indicates high attention" appears in multiple figure captions; consider stating it once in the main text and referring to it in captions.
  3. [References] Reference 22 includes several authors of this paper. While this is a legitimate related work, the self-citation is noticeable; please ensure the citation is necessary and not inflated by self-reference.
  4. [Methods, Model Implementation and Training] The split description would benefit from specifying the random seed and whether the stratification included any additional variables beyond the label (e.g., patient, scanner), which would help reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central claims are empirical benchmarks against external public datasets and standard baselines.

full rationale

I walked the paper's derivation chain and found no step in which a prediction is equivalent by construction to an input, a fitted parameter is renamed as a prediction, or a load-bearing premise rests on a self-citation. The main claim, MST versus 3D ResNet AUC, is a direct empirical comparison on three public datasets with a standard baseline architecture. DINOv2 is an externally pretrained model, and the MST architecture is described and ablated rather than derived from its own output. The explainability evaluation compares model-generated saliency maps against radiologist-annotated lesion locations; this is an external benchmark, not a definitional reduction. The only overlapping-author citation in the introduction (reference 22) supports the general transferability of self-supervised features and is not load-bearing for the paper's specific results. Ablation experiments (Tables 2 and 3) further test components empirically. Possible concerns such as patient-level data leakage would affect validity, not circularity. Therefore, no circular step is present.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claim requires no new physical entities. The free parameters are standard training hyperparameters and input size choices; none are fitted to the test set. The main domain assumptions are the validity of the public dataset labels and the sufficiency of 32-slice crops.

free parameters (6)
  • Learning rate = 1e-4 for 3D ResNet, 1e-5 for MST-ResNet, 1e-6 for MST-DINOv2
    Chosen by hand per architecture; not fitted to test data, but affects performance.
  • Input crop size = 224x224x32 voxels (breast/chest), 150x150x32 resampled to 224x224x32 (knee)
    Fixed input volume dimensions; assumes 32 slices are sufficient for diagnosis.
  • Number of transformer layers = 1
    Single transformer encoder layer used for slice aggregation; chosen without extensive search.
  • Attention heads and feed-forward channels = 12 or 16 heads; 384, 512, or 768 channels depending on encoder
    Set based on the image encoder variant.
  • Weight decay = 1e-2
    AdamW hyperparameter.
  • Early stopping patience = 50 epochs
    Training termination criterion.
assumptions (4)
  • standard math Backpropagation and gradient-based optimization for deep networks
    Used throughout training.
  • domain assumption DINOv2 features pre-trained on natural images transfer to medical slices after fine-tuning
    Core hypothesis; not proven but empirically tested.
  • domain assumption LIDC-IDRI nodule malignancy rating thresholds (ratings 1-2 benign, 4-5 malignant) are valid ground truth
    Methods section 'Data Collection and Preprocessing'.
  • ad hoc to paper A 32-slice crop around the known lesion or relevant anatomy preserves diagnostic information
    Methods: images cropped to 224x224x32. This choice may inflate performance because the model is given the lesion region.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Medical Slice Transformer: Improved Diagnosis and Explainability on 3D Medical Images with DINOv2." pith.science (2026). https://pith.science/paper/IPNYRJN3

@misc{pith2026241115802,
  author       = {Pith},
  title        = {Pith review of: Medical Slice Transformer: Improved Diagnosis and Explainability on 3D Medical Images with DINOv2},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IPNYRJN3}},
  note         = {Machine review of arXiv:2411.15802}
}
abstract

MRI and CT are essential clinical cross-sectional imaging techniques for diagnosing complex conditions. However, large 3D datasets with annotations for deep learning are scarce. While methods like DINOv2 are encouraging for 2D image analysis, these methods have not been applied to 3D medical images. Furthermore, deep learning models often lack explainability due to their "black-box" nature. This study aims to extend 2D self-supervised models, specifically DINOv2, to 3D medical imaging while evaluating their potential for explainable outcomes. We introduce the Medical Slice Transformer (MST) framework to adapt 2D self-supervised models for 3D medical image analysis. MST combines a Transformer architecture with a 2D feature extractor, i.e., DINOv2. We evaluate its diagnostic performance against a 3D convolutional neural network (3D ResNet) across three clinical datasets: breast MRI (651 patients), chest CT (722 patients), and knee MRI (1199 patients). Both methods were tested for diagnosing breast cancer, predicting lung nodule dignity, and detecting meniscus tears. Diagnostic performance was assessed by calculating the Area Under the Receiver Operating Characteristic Curve (AUC). Explainability was evaluated through a radiologist's qualitative comparison of saliency maps based on slice and lesion correctness. P-values were calculated using Delong's test. MST achieved higher AUC values compared to ResNet across all three datasets: breast (0.94$\pm$0.01 vs. 0.91$\pm$0.02, P=0.02), chest (0.95$\pm$0.01 vs. 0.92$\pm$0.02, P=0.13), and knee (0.85$\pm$0.04 vs. 0.69$\pm$0.05, P=0.001). Saliency maps were consistently more precise and anatomically correct for MST than for ResNet. Self-supervised 2D models like DINOv2 can be effectively adapted for 3D medical imaging using MST, offering enhanced diagnostic accuracy and explainability compared to convolutional neural networks.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 2 citations worldwide. Full citation record

  1. Forecasting Continuous Non-Conservative Dynamical Systems in SO(3)

    cs.CV 2025-08 unverdicted novelty 4.0 of 10

    A neural controlled differential equation model with smooth path filtering is proposed for extrapolating 3D rotation trajectories under non-conservative, noisy conditions.

Reference graph

Works this paper leans on

44 extracted references · 33 canonical work pages · cited by 1 Pith paper

  1. [1]

    Lång K, Josefsson V, Larsson AM, Larsson S, Högberg C, Sartor H, et al. Artificial intelligence- supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non- inferiority, single-blinded, screening accuracy study. The Lancet Onco...

    work page 2023

  2. [2]

    Artificial Intelligence in Fracture Detection: A Systematic Review and Meta-Analysis

    Kuo RYL, Harrison C, Curran TA, Jones B, Freethy A, Cussons D, et al. Artificial Intelligence in Fracture Detection: A Systematic Review and Meta-Analysis. Radiology. 2022 Jul;304(1):50–62

    work page 2022

  3. [3]

    Opportunities and obstacles for deep learning in biology and medicine

    Ching T, Himmelstein DS, Beaulieu-Jones BK, Kalinin AA, Do BT, Way GP , et al. Opportunities and obstacles for deep learning in biology and medicine. J R Soc Interface. 2018 Apr;15(141):20170387

    work page 2018

  4. [4]

    Going Deep in Medical Image Analysis: Concepts, Methods, Challenges, and Future Directions

    Altaf F, Islam SMS, Akhtar N, Janjua NK. Going Deep in Medical Image Analysis: Concepts, Methods, Challenges, and Future Directions. IEEE Access. 2019;7:99540–72

    work page 2019

  5. [5]

    Explainable Deep Learning Models in Medical Image Analysis

    Singh A, Sengupta S, Lakshminarayanan V. Explainable Deep Learning Models in Medical Image Analysis. J Imaging. 2020 Jun 20;6(6):52

    work page 2020

  6. [6]

    A Review of Explainable Deep Learning Cancer Detection Models in Medical Imaging

    Gulum MA, Trombley CM, Kantardzic M. A Review of Explainable Deep Learning Cancer Detection Models in Medical Imaging. Applied Sciences. 2021 May 17;11(10):4573

    work page 2021

  7. [7]

    Challenges of Deep Learning in Medical Image Analysis— Improving Explainability and Trust

    Dhar T, Dey N, Borra S, Sherratt RS. Challenges of Deep Learning in Medical Image Analysis— Improving Explainability and Trust. IEEE Trans Technol Soc. 2023 Mar;4(1):68–75

    work page 2023

  8. [8]

    Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization

    Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization. In: 2017 IEEE International Conference on Computer Vision (ICCV) [Internet]. Venice: IEEE; 2017 [cited 2024 Aug 17]. p. 618–26. Available from: http://ieeexplore.ieee.org/document/8237336/

    arXiv 2017

Show all 44 references
  1. [9]

    Understanding the Role of Saliency Maps for Biomarker Research in 3D Medical Imaging Classification

    Shi Y , Wang C, Liu D, Cai W, Cabezas M. Understanding the Role of Saliency Maps for Biomarker Research in 3D Medical Imaging Classification. In: 2023 International Conference on Digital Image 15 Computing: Techniques and Applications (DICTA) [Internet]. Port Macquarie, Austra...

    2023

  2. [10]

    SE3D: A Framework For Saliency Method Evaluation In 3D Imaging [Internet]

    Wiśniewski M, Giulivi L, Boracchi G. SE3D: A Framework For Saliency Method Evaluation In 3D Imaging [Internet]. arXiv; 2024 [cited 2024 Aug 29]. Available from: https://arxiv.org/abs/2405.14584

    2024

  3. [11]

    Attention Is All You Need

    Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, et al. Attention Is All You Need. arXiv:170603762 [cs] [Internet]. 2017 Dec 5 [cited 2021 Mar 29]; Available from: http://arxiv.org/abs/1706.03762

    2017 arXiv

  4. [12]

    An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

    Dosovitskiy A, Beyer L, Kolesnikov A, Weissenborn D, Zhai X, Unterthiner T, et al. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. arXiv:201011929 [cs] [Internet]. 2021 Jun 3 [cited 2022 Mar 28]; Available from: http://arxiv.org/abs/2010.11929

    2021 arXiv

  5. [13]

    Vision transformer to differentiate between benign and malignant slices in 18F-FDG PET/CT

    Nishigaki D, Suzuki Y , Watabe T, Katayama D, Kato H, Wataya T, et al. Vision transformer to differentiate between benign and malignant slices in 18F-FDG PET/CT. Sci Rep. 2024 Apr 9;14(1):8334

    2024

  6. [14]

    Self-evolving vision transformer for chest X- ray diagnosis through knowledge distillation

    Park S, Kim G, Oh Y , Seo JB, Lee SM, Kim JH, et al. Self-evolving vision transformer for chest X- ray diagnosis through knowledge distillation. Nat Commun. 2022 Jul 4;13(1):3848

    2022

  7. [15]

    Attention-based Saliency Maps Improve Interpretability of Pneumothorax Classification

    Wollek A, Graf R, Čečatka S, Fink N, Willem T, Sabel BO, et al. Attention-based Saliency Maps Improve Interpretability of Pneumothorax Classification. Radiology: Artificial Intelligence. 2023 Mar 1;5(2):e220187

    2023

  8. [16]

    An Overview of Deep Semi-Supervised Learning [Internet]

    Ouali Y , Hudelot C, Tami M. An Overview of Deep Semi-Supervised Learning [Internet]. arXiv; 2020 [cited 2024 Aug 29]. Available from: https://arxiv.org/abs/2006.05278

    2020 arXiv

  9. [17]

    DINOv2: Learning Robust Visual Features without Supervision [Internet]

    Oquab M, Darcet T, Moutakanni T, Vo H, Szafraniec M, Khalidov V, et al. DINOv2: Learning Robust Visual Features without Supervision [Internet]. arXiv; 2023 [cited 2024 Aug 16]. Available from: https://arxiv.org/abs/2304.07193

    2023 arXiv

  10. [18]

    Are Natural Domain Foundation Models Useful for Medical Image Classification? In: 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) [Internet]

    Huix JP , Ganeshan AR, Haslum JF, Söderberg M, Matsoukas C, Smith K. Are Natural Domain Foundation Models Useful for Medical Image Classification? In: 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) [Internet]. Waikoloa, HI, USA: IEEE; 2024 [cited 202...

    2024

  11. [19]

    Comparative Analysis of ImageNet Pre- Trained Deep Learning Models and DINOv2 in Medical Imaging Classification [Internet]

    Huang Y , Zou J, Meng L, Yue X, Zhao Q, Li J, et al. Comparative Analysis of ImageNet Pre- Trained Deep Learning Models and DINOv2 in Medical Imaging Classification [Internet]. arXiv; 2024 [cited 2024 Aug 27]. Available from: https://arxiv.org/abs/2402.07595

    2024 arXiv

  12. [20]

    Computer Vision Foundation Models in Endoscopy: Proof of Concept in Oropharyngeal Cancer

    Paderno A, Rau A, Bedi N, Bossi P , Mercante G, Piazza C, et al. Computer Vision Foundation Models in Endoscopy: Proof of Concept in Oropharyngeal Cancer. The Laryngoscope. 2024 Jun 8;lary.31534

    2024

  13. [21]

    DINO-CXR: A self supervised method based on vision transformer for chest X-ray classification [Internet]

    Shakouri M, Iranmanesh F, Eftekhari M. DINO-CXR: A self supervised method based on vision transformer for chest X-ray classification [Internet]. arXiv; 2023 [cited 2024 Jul 1]. Available from: https://arxiv.org/abs/2308.00475

    2023 arXiv

  14. [22]

    Enhancing diagnostic deep learning via self-supervised pretraining on large-scale, unlabeled non-medical images

    Tayebi Arasteh S, Misera L, Kather JN, Truhn D, Nebelung S. Enhancing diagnostic deep learning via self-supervised pretraining on large-scale, unlabeled non-medical images. Eur Radiol Exp. 2024 Feb 8;8(1):10. 16

    2024

  15. [23]

    How Transferable Are Self-supervised Features in Medical Image Classification Tasks? 2021 [cited 2024 Jul 1]; Available from: https://arxiv.org/abs/2108.10048

    Truong T, Mohammadi S, Lenga M. How Transferable Are Self-supervised Features in Medical Image Classification Tasks? 2021 [cited 2024 Jul 1]; Available from: https://arxiv.org/abs/2108.10048

    2021 arXiv

  16. [24]

    Self-Supervision for Medical Image Classification: State-of-the-Art Performance with ~100 Labeled Training Samples per Class

    Nielsen M, Wenderoth L, Sentker T, Werner R. Self-Supervision for Medical Image Classification: State-of-the-Art Performance with ~100 Labeled Training Samples per Class. Bioengineering. 2023 Jul 28;10(8):895

    2023

  17. [25]

    Dynamic contrast-enhanced magnetic resonance images of breast cancer patients with tumor locations [Internet]

    Saha A, Harowicz MR, Grimm LJ, Weng J, Cain EH, Kim CE, et al. Dynamic contrast-enhanced magnetic resonance images of breast cancer patients with tumor locations [Internet]. The Cancer Imaging Archive; 2022 [cited 2024 Oct 6]. Available from: https://www.cancerimagingarchive.n...

    2022

  18. [26]

    Data From LIDC-IDRI [Internet]

    Armato III SG, McLennan G, Bidaut L, McNitt-Gray MF , Meyer CR, Reeves AP , et al. Data From LIDC-IDRI [Internet]. The Cancer Imaging Archive; 2015 [cited 2024 Oct 6]. Available from: https://www.cancerimagingarchive.net/collection/lidc-idri/

    2015

  19. [27]

    Deep-learning-assisted diagnosis for knee magnetic resonance imaging: Development and retrospective validation of MRNet

    Bien N, Rajpurkar P , Ball RL, Irvin J, Park A, Jones E, et al. Deep-learning-assisted diagnosis for knee magnetic resonance imaging: Development and retrospective validation of MRNet. Saria S, editor. PLoS Med. 2018 Nov 27;15(11):e1002699

    2018

  20. [28]

    Deep Residual Learning for Image Recognition

    He K, Zhang X, Ren S, Sun J. Deep Residual Learning for Image Recognition. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) [Internet]. Las Vegas, NV, USA: IEEE; 2016 [cited 2023 Jun 4]. p. 770–8. Available from: http://ieeexplore.ieee.org/document/7780459/

    2016

  21. [29]

    Vision Transformers Need Registers [Internet]

    Darcet T, Oquab M, Mairal J, Bojanowski P . Vision Transformers Need Registers [Internet]. arXiv; 2023 [cited 2024 Aug 16]. Available from: https://arxiv.org/abs/2309.16588

    2023 arXiv

  22. [30]

    Medical Transformer: Universal Encoder for 3-D Brain MRI Analysis

    Jun E, Jeong S, Heo DW, Suk HI. Medical Transformer: Universal Encoder for 3-D Brain MRI Analysis. IEEE Trans Neural Netw Learning Syst. 2023;1–11

    2023

  23. [31]

    Joint transformer architecture in brain 3D MRI classification: its application in Alzheimer’s disease classification

    Alp S, Akan T, Bhuiyan MdS, Disbrow EA, Conrad SA, Vanchiere JA, et al. Joint transformer architecture in brain 3D MRI classification: its application in Alzheimer’s disease classification. Sci Rep. 2024 Apr 18;14(1):8996

    2024

  24. [32]

    M3T: three-dimensional Medical image classifier using Multi-plane and Multi-slice Transformer

    Jang J, Hwang D. M3T: three-dimensional Medical image classifier using Multi-plane and Multi-slice Transformer. In: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) [Internet]. New Orleans, LA, USA: IEEE; 2022 [cited 2024 Sep 21]. p. 20686–97. Availab...

    2022

  25. [33]

    Visual Transformers and Convolutional Neural Networks for Disease Classification on Radiographs: A Comparison of Performance, Sample Efficiency, and Hidden Stratification

    Murphy ZR, Venkatesh K, Sulam J, Yi PH. Visual Transformers and Convolutional Neural Networks for Disease Classification on Radiographs: A Comparison of Performance, Sample Efficiency, and Hidden Stratification. Radiology: Artificial Intelligence. 2022 Nov 1;4(6):e220012

    2022

  26. [34]

    Towards Evaluating Explanations of Vision Transformers for Medical Imaging

    Komorowski P , Baniecki H, Biecek P . Towards Evaluating Explanations of Vision Transformers for Medical Imaging. In: 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) [Internet]. Vancouver, BC, Canada: IEEE; 2023 [cited 2024 Sep 26]. p. 3726–

    2023

  27. [35]

    Available from: https://ieeexplore.ieee.org/document/10208407/

  28. [36]

    A machine learning approach to radiogenomics of breast cancer: a study of 922 subjects and 529 DCE-MRI features

    Saha A, Harowicz MR, Grimm LJ, Kim CE, Ghate SV, Walsh R, et al. A machine learning approach to radiogenomics of breast cancer: a study of 922 subjects and 529 DCE-MRI features. Br J Cancer. 2018 Aug;119(4):508–16. 17

    2018

  29. [37]

    Hancock MC, Magnan JF. Lung nodule malignancy classification using only radiologist- quantified image features as inputs to statistical learning algorithms: probing the Lung Image Database Consortium dataset with two statistical learning methods. J Med Imag. 2016 Dec 8;3(4):044504

    2016

  30. [38]

    MONAI: An open-source framework for deep learning in healthcare

    Cardoso MJ, Li W, Brown R, Ma N, Kerfoot E, Wang Y , et al. MONAI: An open-source framework for deep learning in healthcare. 2022 [cited 2023 Jul 13]; Available from: https://arxiv.org/abs/2211.02701

    2022 arXiv

  31. [39]

    TorchIO: A Python library for efficient loading, preprocessing, augmentation and patch-based sampling of medical images in deep learning

    Pérez-García F, Sparks R, Ourselin S. TorchIO: A Python library for efficient loading, preprocessing, augmentation and patch-based sampling of medical images in deep learning. 2020 [cited 2023 Apr 13]; Available from: https://arxiv.org/abs/2003.04696

    2020 arXiv

  32. [40]

    Decoupled Weight Decay Regularization

    Loshchilov I, Hutter F. Decoupled Weight Decay Regularization. 2017 [cited 2024 Jan 10]; Available from: https://arxiv.org/abs/1711.05101

    2017 arXiv

  33. [41]

    Scikit-learn: Machine learning in Python

    Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research. 2011;12:2825–30

    2011

  34. [42]

    Comparing the Areas under Two or More Correlated Receiver Operating Characteristic Curves: A Nonparametric Approach

    DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the Areas under Two or More Correlated Receiver Operating Characteristic Curves: A Nonparametric Approach. Biometrics. 1988 Sep;44(3):837

    1988

  35. [43]

    Fast Implementation of DeLong’s Algorithm for Comparing the Areas Under Correlated Receiver Operating Characteristic Curves

    Sun X, Xu W. Fast Implementation of DeLong’s Algorithm for Comparing the Areas Under Correlated Receiver Operating Characteristic Curves. IEEE Signal Process Lett. 2014 Nov;21(11):1389–93

    2014

  36. [44]

    Does the saliency map highlight the slice(s) containing the lesion?

    Virtanen P , Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, et al. SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat Methods. 2020 Mar 2;17(3):261– 72. Acknowledgements • The data used in this publication was managed using the research data...

    2020

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.