pith. sign in

arxiv: 2605.12570 · v1 · pith:SQMGTQVJnew · submitted 2026-05-12 · 💻 cs.CV

M3Net: A Macro-to-Meso-to-Micro Clinical-inspired Hierarchical 3D Network for Pulmonary Nodule Classification

Jinyue Li , Yuzhou Yu , Jingjing Yang , Meng Fu , Yani Zhang , Shuyao He , Dianlong Ge , Xin Ning
show 2 more authors
Yannan Chu Qiankun Li
This is my paper

Pith reviewed 2026-05-14 20:52 UTC · model grok-4.3

classification 💻 cs.CV
keywords pulmonary nodule classification3D hierarchical networkmulti-scale CT analysislung cancer screeningmutual information maximizationclinical-inspired modelbenign malignant distinction
0
0 comments X

The pith

M3Net classifies pulmonary nodules more accurately by processing CT scans in a radiologist-inspired macro-to-micro hierarchy.

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

The paper presents M3Net, a 3D network designed to classify benign and malignant pulmonary nodules in CT scans by replicating the hierarchical diagnostic approach used by radiologists. It constructs inputs at multiple scales, from fine nodule details to local and global contexts, and maintains consistency across these scales using latent space projections and mutual information maximization. This addresses the challenges of nodule heterogeneity and aims to move beyond opaque deep learning models toward more clinically aligned AI. A reader would care because better classification can aid early lung cancer detection and make AI tools more trustworthy in medical settings.

Core claim

The central discovery is that a hierarchical 3D network with progressive multi-scale inputs from fine-grained structures to global relationships, combined with scale-specific encoders and cross-scale consistency enforced by latent space projection and mutual information maximization, achieves state-of-the-art classification accuracies of 86.96% on LIDC-IDRI and 84.24% on USTC-FHLN datasets, surpassing the best baseline by 3.26% and 2.17%.

What carries the argument

Progressive multi-scale input construction using scale-specific encoders, with latent space projection and mutual information maximization to ensure cross-scale semantic consistency.

If this is right

  • The approach yields more accurate benign-malignant distinctions in pulmonary nodules.
  • It offers improved robustness on both public benchmarks and clinical data.
  • The clinical workflow inspiration enhances potential for explainable AI in radiology.
  • Such networks can better handle the multi-scale and heterogeneous characteristics of nodules.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar multi-scale hierarchical designs could apply to other 3D medical imaging tasks like tumor detection in different organs.
  • The mutual information component may help mitigate issues with limited labeled medical data.
  • Deployment in varied clinical environments could test the method's adaptability to different CT protocols.

Load-bearing premise

That integrating multi-scale inputs with cross-scale consistency mechanisms leads to clinically meaningful improvements in nodule classification that hold beyond the evaluated datasets.

What would settle it

If a new dataset with greater diversity in nodule types, patient demographics, or scanner variations shows no accuracy improvement over baselines, the effectiveness claim would be challenged.

Figures

Figures reproduced from arXiv: 2605.12570 by Dianlong Ge, Jingjing Yang, Jinyue Li, Meng Fu, Qiankun Li, Shuyao He, Xin Ning, Yani Zhang, Yannan Chu, Yuzhou Yu.

Figure 1
Figure 1. Figure 1: Overview of M3Net. (a) Radiologists’ diagnostic workflow motivating the two-stage framework. (b) The proposed method exploits neighborhood information of lung nodules through three independently fine-tuned classifiers with complementary context. global screening to localize suspicious regions, (2) bound￾ary inspection to evaluate margin sharpness and lobulation, and (3) fine-grained internal assessment for… view at source ↗
Figure 2
Figure 2. Figure 2: Example images from the LIDC-IDRI and USTC-FHLN datasets [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (a) Confusion matrices for pulmonary nodule classification on LIDC-IDRI dataset; (b) Accuracy, F1-score, Precision, and Recall comparison on LIDC-IDRI and USTC-FHLN datasets; (c) Specificity, Balanced Accuracy, ROC AUC, and PR AUC visualization on LIDC-IDRI and USTC-FHLN datasets. Performance comparison of various methods for benign-malignant pulmonary nodule classification. 473 39 78 85 Predicted class be… view at source ↗
Figure 4
Figure 4. Figure 4: The comparison results illustrate that our method consistently outperforms competing methods by effectively correcting a substantial portion of misclassified and incorrectly predicted samples. In particular, the proposed approach demonstrates a stronger ability to recover correct predictions in challenging cases [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Grad-CAM visualizations for two representative cases. For each case, the original CT slice and the corresponding attention maps generated by different models are shown. Warmer colors indicate regions contributing more to the model prediction. The visualizations highlight how the models focus on the nodule regions and surrounding contextual structures during classification. Displayed slices correspond to ea… view at source ↗
read the original abstract

The accurate classification of benign and malignant pulmonary nodules in CT scans is critical for early lung cancer screening, yet remains challenging due to the multi-scale and heterogeneous nature of pulmonary nodules. While deep learning offers potential for auxiliary diagnosis, most existing models act as "black boxes", lacking the transparency and explainability required for trustworthy clinical integration. To address this issue, we propose M3Net, a novel 3D network for pulmonary nodule classification inspired by the hierarchical diagnostic workflow of radiologists, which integrates multi-scale contextual information from fine-grained structures to global anatomical relationships. Our framework constructs a progressive multi-scale input, from fine-grained nodule structures to local semantics and global spatial relationships. M3Net employs scale-specific encoders and ensures cross-scale semantic consistency through latent space projection and mutual information maximization. Extensive experiments on the public LIDC-IDRI dataset and a self-collected clinical dataset (USTC-FHLN) demonstrate that our method achieves state-of-the-art performance, with accuracies of 86.96% and 84.24% respectively, outperforming the best baseline by 3.26% and 2.17%. The results validate that M3Net provides a more robust and clinically relevant solution for pulmonary nodule classification. The code is available at https://github.com/jylEcho/M3-Net.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes M3Net, a 3D hierarchical network for benign/malignant pulmonary nodule classification in CT scans, inspired by radiologists' macro-to-meso-to-micro diagnostic workflow. It builds progressive multi-scale inputs, employs scale-specific encoders, and enforces cross-scale consistency via latent-space projection and mutual information maximization. On LIDC-IDRI it reports 86.96% accuracy and on the private USTC-FHLN dataset 84.24% accuracy, outperforming the strongest baseline by 3.26% and 2.17% respectively; code is released publicly.

Significance. If the accuracy gains are shown to be statistically reliable and causally linked to the hierarchical mechanism, the work would supply a clinically aligned, more interpretable alternative to black-box 3D CNNs for early lung-cancer screening. Public code release supports reproducibility and follow-up studies.

major comments (2)
  1. [Abstract / Experiments] Abstract and experimental section: point accuracies (86.96%, 84.24%) and claimed margins (3.26%, 2.17%) are supplied without standard deviations, p-values, multiple-seed results, or dataset-split details, so the central SOTA claim cannot be verified.
  2. [Method / Experiments] Method and experiments: no ablation tables isolate the contribution of progressive multi-scale input construction, latent-space projection, or the mutual-information term; without them the attribution of the reported lifts to the clinical-inspired hierarchy remains untested.
minor comments (1)
  1. [Experiments] Clarify whether the same training schedule, optimizer, and data-augmentation pipeline were used for all baselines to ensure fair comparison.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which highlight important aspects for strengthening the statistical rigor and component analysis in our work. We will revise the manuscript accordingly to address these points.

read point-by-point responses

  1. Referee: [Abstract / Experiments] Abstract and experimental section: point accuracies (86.96%, 84.24%) and claimed margins (3.26%, 2.17%) are supplied without standard deviations, p-values, multiple-seed results, or dataset-split details, so the central SOTA claim cannot be verified.

    Authors: We agree that the current version lacks these statistical details, limiting verification of the SOTA claims. In the revised manuscript, we will report mean accuracies with standard deviations computed over multiple random seeds (e.g., 5 independent runs), include p-values from paired statistical tests for the performance margins against baselines, and provide explicit details on dataset splits (including train/validation/test ratios and any cross-validation procedure used on LIDC-IDRI and USTC-FHLN). revision: yes

  2. Referee: [Method / Experiments] Method and experiments: no ablation tables isolate the contribution of progressive multi-scale input construction, latent-space projection, or the mutual-information term; without them the attribution of the reported lifts to the clinical-inspired hierarchy remains untested.

    Authors: We concur that ablation studies are required to isolate and attribute the contributions of each component. In the revised version, we will add comprehensive ablation tables in the experiments section. These will include performance results for variants that disable progressive multi-scale input construction, remove the latent-space projection, and omit the mutual information maximization term, allowing direct assessment of their individual impacts on the observed accuracy gains. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The provided manuscript text (abstract and description) contains no equations, derivations, or load-bearing steps that reduce any claimed prediction or result to its inputs by construction. The architecture is described as a progressive multi-scale 3D network with scale-specific encoders, latent projection, and mutual information maximization, but these are presented as design choices rather than self-referential definitions. Performance claims consist of empirical accuracies (86.96% on LIDC-IDRI, 84.24% on USTC-FHLN) measured against external baselines on held-out data; no fitted parameters are renamed as predictions, no uniqueness theorems are invoked via self-citation, and no ansatz or renaming of known results appears. The central claims rest on experimental validation, making the derivation self-contained with no circular reductions.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on empirical results from a trained 3D neural network whose parameters are fitted to the two datasets; the design assumes that radiologist-style hierarchical viewing can be captured by multi-scale encoders with consistency constraints.

free parameters (1)
  • network weights and hyperparameters
    All convolutional and projection layer parameters are learned from the training data; no specific count or values are given in the abstract.
axioms (1)
  • domain assumption Hierarchical multi-scale processing with cross-scale consistency improves classification of heterogeneous pulmonary nodules
    This is the core modeling assumption that justifies the macro-to-meso-to-micro architecture.

pith-pipeline@v0.9.0 · 5570 in / 1284 out tokens · 43392 ms · 2026-05-14T20:52:29.831615+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

61 extracted references · 61 canonical work pages · 2 internal anchors

  1. [1]

    , 2023. Expanding role of advanced image analysis in CT-detected indeterminate pulmonary nodules and early lung cancer characteri- zation, author=Prosper, Ashley Elizabeth and Kammer, Michael N and Maldonado, Fabien and Aberle, Denise R and Hsu, William. Radiology 309, e222904

    work page 2023

  2. [2]

    , 2024. Deep learning for malignancy risk estimation of incidental sub-centimeter pulmonary nodules on CT images, author=Zhang, Rui and Wei, Ying and Wang, Denian and Chen, Bojiang and Sun, Huaiqiang and Lei, Yi and Zhou, Qing and Luo, Zhuang and Jiang, Li and Qiu, Rong and others. European Radiology 34, 4218–4229

    work page 2024

  3. [3]

    , 2024. Lung nodule malignancy classification with associated pulmonary fibrosis using 3D attention-gated convolutional network with CT scans, author=Liu, Yucheng and Hsu, Hao Yun and Lin, Tiffany and Peng, Boyu and Saqi, Anjali and Salvatore, Mary M and Jambawalikar, Sachin. Journal of Translational Medicine 22, 51

    work page 2024

  4. [4]

    Artificialneuralnetworkvalidationofmhdnaturalbioconvectionina squareenclosure:entropicanalysisandoptimization.ActaMechanica Sinica 41, 724507

    Alsedais, N., Mansour, M.A., Aly, A.M., Abdelsalam, S.I., 2025. Artificialneuralnetworkvalidationofmhdnaturalbioconvectionina squareenclosure:entropicanalysisandoptimization.ActaMechanica Sinica 41, 724507

    work page 2025

  5. [5]

    The lung image database consortium (LIDC) and image database resource initiative (IDRI): a completed referencedatabaseoflungnodulesonCTscans

    Armato III, S.G., McLennan, G., Bidaut, L., McNitt-Gray, M.F., Meyer, C.R., Reeves, A.P., Zhao, B., Aberle, D.R., Henschke, C.I., Hoffman, E.A., et al., 2011. The lung image database consortium (LIDC) and image database resource initiative (IDRI): a completed referencedatabaseoflungnodulesonCTscans. MedicalPhysics38, 915–931

    work page 2011

  6. [6]

    Are we justified attributing a mistake in diagnosis to an ai diagnostic system? AI and Ethics 3, 567–584

    Babushkina, D., 2023. Are we justified attributing a mistake in diagnosis to an ai diagnostic system? AI and Ethics 3, 567–584

    work page 2023

  7. [7]

    M3d: Advancing 3d medical image analysis with multi-modal large language models,

    Bai,F.,Du,Y.,Huang,T.,Meng,M.Q.H.,Zhao,B.,2024. M3d:Ad- vancing 3d medical image analysis with multi-modal large language models. arXiv preprint arXiv:2404.00578

    work page arXiv 2024

  8. [8]

    Clinician-informed xai evaluation checklist with metrics (clix-m) for ai-powered clinical decision support systems

    Brankovic, A., Cook, D., Rahman, J., Delaforce, A., Li, J., Magrabi, F., Cabitza, F., Coiera, E., Bradford, D., 2025. Clinician-informed xai evaluation checklist with metrics (clix-m) for ai-powered clinical decision support systems. npj Digital Medicine 8, 364

    work page 2025

  9. [9]

    Reproducible and interpretable spiculation quantification for lung cancer screening

    Choi,W.,Nadeem,S.,Alam,S.R.,Deasy,J.O.,Tannenbaum,A.,Lu, W., 2021. Reproducible and interpretable spiculation quantification for lung cancer screening. Computer methods and programs in biomedicine 200, 105839

    work page 2021

  10. [10]

    Concept-based explainable malignancy scoring on pulmonary nodules in ct images

    Dumaev, R.I., Molodyakov, S.A., Utkin, L.V., 2024. Concept-based explainable malignancy scoring on pulmonary nodules in ct images. arXiv preprint arXiv:2405.17483

    work page arXiv 2024

  11. [11]

    Arti- ficial intelligence-driven fvm-ann model for entropy analysis of mhd natural bioconvection in nanofluid-filled porous cavities

    Elsedais, N., Mansour, M.A., Aly, A.M., Abdelsalam, S., 2024. Arti- ficial intelligence-driven fvm-ann model for entropy analysis of mhd natural bioconvection in nanofluid-filled porous cavities

    work page 2024

  12. [12]

    EMeR- ALDS: Electronic Medical Record Driven Automated Lung Nodule Detection and Classification in Thoracic CT Images

    Eman, H., Shaukat, F., Zafar, M.H., Anwar, S.M., 2025. EMeR- ALDS: Electronic Medical Record Driven Automated Lung Nodule Detection and Classification in Thoracic CT Images. arXiv preprint arXiv:2509.11714

    work page arXiv 2025

  13. [13]

    Semi-supervised adversarial learning for improving the diagnosis of pulmonary nodules

    Fu, Y., Xue, P., Xiao, T., Zhang, Z., Zhang, Y., Dong, E., 2022. Semi-supervised adversarial learning for improving the diagnosis of pulmonary nodules. IEEE Journal of Biomedical and Health Informatics 27, 109–120

    work page 2022

  14. [14]

    Combining multistaged filters and modi- fied segmentation network for improving lung nodules classification

    Gunawan, R., Tran, Y., Zheng, J., Nguyen, H., Carrigan, A., Mills, M.K., Chai, R., 2024. Combining multistaged filters and modi- fied segmentation network for improving lung nodules classification. IEEE Journal of Biomedical and Health Informatics 28, 5519–5527

    work page 2024

  15. [15]

    Texture and radiomics inspired data-driven cancerous lung nodules severity classification

    Gupta, H., Singh, H., Kumar, A., 2024. Texture and radiomics inspired data-driven cancerous lung nodules severity classification. Biomedical Signal Processing and Control 88, 105543

    work page 2024

  16. [16]

    Accurateclassificationofpulmonary nodulesbyacombinedmodelofclinical,imaging,andcell-freeDNA methylationbiomarkers:amodeldevelopmentandexternalvalidation study

    He, J., Wang, B., Tao, J., Liu, Q., Peng, M., Xiong, S., Li, J., Cheng, B.,Li,C.,Jiang,S.,etal.,2023. Accurateclassificationofpulmonary nodulesbyacombinedmodelofclinical,imaging,andcell-freeDNA methylationbiomarkers:amodeldevelopmentandexternalvalidation study. The Lancet Digital Health 5, e647–e656

    work page 2023

  17. [17]

    Deep residual learning for image recognition, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    He, K., Zhang, X., Ren, S., Sun, J., 2016. Deep residual learning for image recognition, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778

    work page 2016

  18. [18]

    An ISHAP-based interpretation-model-guided classification method for malignant pul- monary nodule

    He, W., Li, B., Liao, R., Mo, H., Tian, L., 2022. An ISHAP-based interpretation-model-guided classification method for malignant pul- monary nodule. Knowledge-Based Systems 237, 107778

    work page 2022

  19. [19]

    Learning efficient, explainableanddiscriminativerepresentationsforpulmonarynodules classification

    Jiang, H., Shen, F., Gao, F., Han, W., 2021. Learning efficient, explainableanddiscriminativerepresentationsforpulmonarynodules classification. Pattern Recognition 113, 107825

    work page 2021

  20. [20]

    Guidelinesandeval- uation of clinical explainable ai in medical image analysis

    Jin,W.,Li,X.,Fatehi,M.,Hamarneh,G.,2023. Guidelinesandeval- uation of clinical explainable ai in medical image analysis. Medical image analysis 84, 102684

    work page 2023

  21. [21]

    Liu, Z., Mao, H., Wu, C.Y., Feichtenhofer, C., Darrell, T., Xie, S.,

    work page

  22. [22]

    A convnet for the 2020s, in: Proceedings of the IEEE/CVF ConferenceonComputerVisionandPatternRecognition,pp.11976– 11986

    work page

  23. [23]

    Explainable artificial intelligence (xai) 2.0: A manifesto of open challenges and interdisciplinary research directions

    Longo, L., Brcic, M., Cabitza, F., Choi, J., Confalonieri, R., Del Ser, J., Guidotti, R., Hayashi, Y., Herrera, F., Holzinger, A., et al., 2024. Explainable artificial intelligence (xai) 2.0: A manifesto of open challenges and interdisciplinary research directions. Information Fusion 106, 102301

    work page 2024

  24. [24]

    A novel benign and malignant classification model for lung nodules based on multi- scale interleaved fusion integrated network

    Lv, E., Kang, X., Wen, P., Tian, J., Zhang, M., 2024. A novel benign and malignant classification model for lung nodules based on multi- scale interleaved fusion integrated network. Scientific Reports 14, 27506

    work page 2024

  25. [25]

    Memory- augmented capsule network for adaptable lung nodule classification

    Mobiny, A., Yuan, P., Cicalese, P.A., Moulik, S.K., Garg, N., Wu, C.C.,Wong,K.,Wong,S.T.,He,T.C.,Nguyen,H.V.,2021. Memory- augmented capsule network for adaptable lung nodule classification. IEEE Transactions on Medical Imaging 40, 2869–2879

    work page 2021

  26. [26]

    Multilayer outperforms single-layer slide scanning in ai-based classification of whole slide images with low-burden acid-fast mycobacteria (afb)

    Nurzynska, K., Li, D., Walts, A.E., Gertych, A., 2023. Multilayer outperforms single-layer slide scanning in ai-based classification of whole slide images with low-burden acid-fast mycobacteria (afb). Computer Methods and Programs in Biomedicine 234, 107518

    work page 2023

  27. [27]

    EfficacyofultrashortechotimepulmonaryMRIforlung noduledetectionandlung-RADSclassification

    Ohno, Y., Takenaka, D., Yoshikawa, T., Yui, M., Koyama, H., Ya- mamoto,K.,Hamabuchi,N.,Shigemura,C.,Watanabe,A.,Ueda,T., etal.,2022. EfficacyofultrashortechotimepulmonaryMRIforlung noduledetectionandlung-RADSclassification. Radiology302,697– 706

    work page 2022

  28. [28]

    DINOv2: Learning Robust Visual Features without Supervision

    Oquab,M.,Darcet,T.,Moutakanni,T.,Vo,H.,Szafraniec,M.,Khali- dov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., Assran, M., Ballas, N., Galuba, W., Howes, R., Huang, P.Y., Li, S.W., Misra, I., Rabbat, M., Sharma, V., Synnaeve, G., Xu, H., Jegou, H., Mairal, J., Labatut, P., Joulin, A., Bojanowski, P., 2023. DINOv2: Learning Robust Visual Features...

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  29. [29]

    ADGAN:Attribute- driven generative adversarial network for synthesis and multiclass classification of pulmonary nodules

    Roy,R.,Mazumdar,S.,Chowdhury,A.S.,2022. ADGAN:Attribute- driven generative adversarial network for synthesis and multiclass classification of pulmonary nodules. IEEE Transactions on Neural Networks and Learning Systems 35, 2484–2495

    work page 2022

  30. [30]

    Efficient pul- monary nodules classification using radiomics and different artificial intelligence strategies

    Saied, M., Raafat, M., Yehia, S., Khalil, M.M., 2023. Efficient pul- monary nodules classification using radiomics and different artificial intelligence strategies. Insights into Imaging 14, 91

    work page 2023

  31. [31]

    An interpretabledeephierarchicalsemanticconvolutionalneuralnetwork for lung nodule malignancy classification

    Shen, S., Han, S.X., Aberle, D.R., Bui, A.A., Hsu, W., 2019. An interpretabledeephierarchicalsemanticconvolutionalneuralnetwork for lung nodule malignancy classification. Expert Systems with Applications 128, 84–95

    work page 2019

  32. [32]

    Multi-crop convolutional neural networks for lung nodulemalignancysuspiciousnessclassification

    Shen, W., Zhou, M., Yang, F., Yu, D., Dong, D., Yang, C., Zang, Y., Tian, J., 2017. Multi-crop convolutional neural networks for lung nodulemalignancysuspiciousnessclassification. PatternRecognition 61, 663–673

    work page 2017

  33. [33]

    Siméoni, O., Vo, H.V., Seitzer, M., Baldassarre, F., Oquab, M., Jose, C., Khalidov, V., Szafraniec, M., Yi, S., Ramamonjisoa, M., Massa, F., Haziza, D., Wehrstedt, L., Wang, J., Darcet, T., Moutakanni, T., Sentana, L., Roberts, C., Vedaldi, A., Tolan, J., Brandt, J., Couprie, C., Mairal, J., Jégou, H., Labatut, P., Bojanowski, P., 2025. DINOv3. arXiv:2508.10104

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  34. [34]

    Streamlinepathologyfoundationmodelbycross-magnificationdistil- lation

    Su, Z., Akbar, A.R., Sajjad, U., Parwani, A.V., Niazi, M.K.K., 2025. Streamlinepathologyfoundationmodelbycross-magnificationdistil- lation. arXiv preprint arXiv:2509.23097

    work page arXiv 2025

  35. [35]

    Tang, H., Zhang, C., Xie, X., 2019. Nodulenet: Decoupled false positive reduction for pulmonary nodule detection and segmenta- tion, in: International Conference on Medical Image Computing and Computer-Assisted Intervention, Springer. pp. 266–274

    work page 2019

  36. [36]

    Automated ab- normalityclassificationofchestradiographsusingdeepconvolutional neural networks

    Tang, Y.X., Tang, Y.B., Peng, Y., Yan, K., Bagheri, M., Redd, B.A., Brandon, C.J., Lu, Z., Han, M., Xiao, J., et al., 2020. Automated ab- normalityclassificationofchestradiographsusingdeepconvolutional neural networks. NPJ Digital Medicine 3, 70

    work page 2020

  37. [37]

    Ethical considerations intheuseofartificialintelligenceandmachinelearninginhealthcare: a comprehensive review

    Tilala, M.H., Chenchala, P.K., Choppadandi, A., Kaur, J., Naguri, S., Saoji, R., Devaguptapu, B., Tilala, M., 2024. Ethical considerations intheuseofartificialintelligenceandmachinelearninginhealthcare: a comprehensive review. Cureus 16

    work page 2024

  38. [38]

    Tong, C., Liang, B., Su, Q., Yu, M., Hu, J., Bashir, A.K., Zheng, Z.,

    work page

  39. [39]

    IEEE Journal on Selected Areas in Communications 39, 574–581

    Pulmonary nodule classification based on heterogeneous fea- tures learning. IEEE Journal on Selected Areas in Communications 39, 574–581

    work page

  40. [40]

    Towards reliable and explainable ai model for pulmonary nodule diagnosis

    Wang, C., Liu, Y., Wang, F., Zhang, C., Wang, Y., Yuan, M., Yang, G., 2024a. Towards reliable and explainable ai model for pulmonary nodule diagnosis. Biomedical Signal Processing and Control 88, 105646

    work page

  41. [41]

    Amulti-objectivesegmentationmethodforchestx-raysbased on collaborative learning from multiple partially annotated datasets

    Wang, H., Zhang, D., Feng, J., Cascone, L., Nappi, M., Wan, S., 2024b. Amulti-objectivesegmentationmethodforchestx-raysbased on collaborative learning from multiple partially annotated datasets. Information Fusion 102, 102016

    work page

  42. [42]

    Explainable machine learning model based on clinical factors for predicting the disappearance of indeterminate pulmonarynodules.Computersinbiologyandmedicine169,107871

    Wang,J.,Sourlos,N.,Heuvelmans,M.,Prokop,M.,Vliegenthart,R., van Ooijen, P., 2024c. Explainable machine learning model based on clinical factors for predicting the disappearance of indeterminate pulmonarynodules.Computersinbiologyandmedicine169,107871

    work page

  43. [43]

    Integrative serum metabolic fingerprints based multi-modal platforms for lung adenocarcinoma early detection and pulmonary nodule classification

    Wang,L.,Zhang,M.,Pan,X.,Zhao,M.,Huang,L.,Hu,X.,Wang,X., Qiao, L., Guo, Q., Xu, W., et al., 2022. Integrative serum metabolic fingerprints based multi-modal platforms for lung adenocarcinoma early detection and pulmonary nodule classification. Advanced Science 9, 2203786

    work page 2022

  44. [44]

    Wang, X., Peng, Y., Lu, L., Lu, Z., Bagheri, M., Summers, R.M.,

    work page

  45. [45]

    2097–2106

    Chestx-ray8: Hospital-scale chest X-ray database and bench- marks on weakly-supervised classification and localization of com- mon thorax diseases, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 2097–2106

    work page 2097

  46. [46]

    Explainable ai-driven iomt fusion: Unravelling techniques, opportunities, and challengeswithexplainableaiinhealthcare

    Wani, N.A., Kumar, R., Bedi, J., Rida, I., et al., 2024. Explainable ai-driven iomt fusion: Unravelling techniques, opportunities, and challengeswithexplainableaiinhealthcare. InformationFusion110, 102472

    work page 2024

  47. [47]

    arXiv preprint arXiv:2006.03677 (2020) Multimodal Learning for Vocal Tract Segmentation 11

    Wu, B., Xu, C., Dai, X., Wan, A., Zhang, P., Yan, Z., Tomizuka, M., Gonzalez, J., Keutzer, K., Vajda, P., 2020. Visual Transformers: Token-based Image Representation and Processing for Computer Vision.arXiv:2006.03677

    work page arXiv 2020

  48. [48]

    Knowledge-basedcollaborativedeeplearningforbenign- malignantlungnoduleclassificationonchestCT

    Xie, Y., Xia, Y., Zhang, J., Song, Y., Feng, D., Fulham, M., Cai, W.,2018a. Knowledge-basedcollaborativedeeplearningforbenign- malignantlungnoduleclassificationonchestCT. IEEETransactions on Medical Imaging 38, 991–1004

    work page

  49. [49]

    Semi-supervised adversarial model forbenign–malignantlungnoduleclassificationonchestCT

    Xie, Y., Zhang, J., Xia, Y., 2019. Semi-supervised adversarial model forbenign–malignantlungnoduleclassificationonchestCT. Medical Image Analysis 57, 237–248

    work page 2019

  50. [50]

    Fusing texture, shape and deep model-learned information at decision level forautomatedclassificationoflungnodulesonchestCT

    Xie, Y., Zhang, J., Xia, Y., Fulham, M., Zhang, Y., 2018b. Fusing texture, shape and deep model-learned information at decision level forautomatedclassificationoflungnodulesonchestCT. Information Fusion 42, 102–110

    work page

  51. [51]

    Pulmonary nodule detection based on model fusion and adaptive false positive reduction

    Xiong, Y., Deng, L., Wang, Y., 2024. Pulmonary nodule detection based on model fusion and adaptive false positive reduction. Expert Systems with Applications 238, 121890

    work page 2024

  52. [52]

    A generalizable 3d framework and model for self-supervised learning in medical imaging

    Xu, T., Hosseini, S., Anderson, C., Rinaldi, A., Krishnan, R.G., Martel, A.L., Goubran, M., 2025. A generalizable 3d framework and model for self-supervised learning in medical imaging. npj Digital Medicine 8, 639

    work page 2025

  53. [53]

    Tdf-net: Trusted dynamic feature fusion network for breast cancer diagnosis using incomplete multimodal ultrasound

    Yan, P., Gong, W., Li, M., Zhang, J., Li, X., Jiang, Y., Luo, H., Zhou, H., 2024. Tdf-net: Trusted dynamic feature fusion network for breast cancer diagnosis using incomplete multimodal ultrasound. Information Fusion 112, 102592

    work page 2024

  54. [54]

    Multi-labelsoftmaxnetworks for pulmonary nodule classification using unbalanced and dependent categories

    Yi,L.,Zhang,L.,Xu,X.,Guo,J.,2022. Multi-labelsoftmaxnetworks for pulmonary nodule classification using unbalanced and dependent categories. IEEE Transactions on Medical Imaging 42, 317–328

    work page 2022

  55. [55]

    Anattentiveandadaptive 3D CNN for automatic pulmonary nodule detection in CT image

    Zhao,D.,Liu,Y.,Yin,H.,Wang,Z.,2023. Anattentiveandadaptive 3D CNN for automatic pulmonary nodule detection in CT image. Expert Systems with Applications 211, 118672

    work page 2023

  56. [56]

    Medical image classification using light-weight CNN with spiking cortical model based attention module

    Zhou, Q., Huang, Z., Ding, M., Zhang, X., 2023. Medical image classification using light-weight CNN with spiking cortical model based attention module. IEEE Journal of Biomedical and Health Informatics 27, 1991–2002

    work page 2023

  57. [57]

    A cascaded multi-stage framework for automatic detection and segmentation of pulmonary nodules in developing countries

    Zhou, Z., Gou, F., Tan, Y., Wu, J., 2022. A cascaded multi-stage framework for automatic detection and segmentation of pulmonary nodules in developing countries. IEEE Journal of Biomedical and Health Informatics 26, 5619–5630

    work page 2022

  58. [58]

    Explainableclassification of benign-malignant pulmonary nodules with neural networks and information bottleneck

    Zhu,H.,Liu,W.,Gao,Z.,Zhang,H.,2023. Explainableclassification of benign-malignant pulmonary nodules with neural networks and information bottleneck. IEEE Transactions on Neural Networks and Learning Systems

    work page 2023

  59. [59]

    Multi-view coupled self-attention network for pulmonary nodules classification, in: Proceedings of the Asian Conference on Computer Vision, pp

    Zhu, Q., Wang, Y., Chu, X., Yang, X., Zhong, W., 2022. Multi-view coupled self-attention network for pulmonary nodules classification, in: Proceedings of the Asian Conference on Computer Vision, pp. 995–1009

    work page 2022

  60. [60]

    Zhu,W.,Liu,C.,Fan,W.,Xie,X.,2018.Deeplung:Deep3Ddualpath netsforautomatedpulmonarynoduledetectionandclassification,in: 2018 IEEE Winter Conference on Applications of Computer Vision (WACV), IEEE. pp. 673–681

    work page 2018

  61. [61]

    Uni- formizingtechniquestoprocessCTscanswith3DCNNsfortubercu- losisprediction,in:InternationalWorkshoponPredictiveIntelligence in Medicine, Springer

    Zunair, H., Rahman, A., Mohammed, N., Cohen, J.P., 2020. Uni- formizingtechniquestoprocessCTscanswith3DCNNsfortubercu- losisprediction,in:InternationalWorkshoponPredictiveIntelligence in Medicine, Springer. pp. 156–168

    work page 2020