Pith. sign in

REVIEW 4 major objections 4 minor 72 references

Feature-interactive Siamese graph encoder-based image analysis to predict STAS from histopathology images in lung cancer

T0 review · 4 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 feature-interactive Siamese graph encoder, VERN, predicts spread through air spaces (STAS) in lung cancer whole-slide images, reporting an internal AUC of 0.9215 and frozen- and paraffin-section test AUCs of 0.8275 and 0.8829.

desk verdict A solid first WSI-level STAS prediction model with genuine external validation, but the internal split needs to be patient-level before the headline AUC is taken at face value. read the letter →

arxiv 2411.15274 v1 pith:IISQVLO5 submitted 2024-11-22 eess.IV cs.AIcs.CV

Liangrui Pan , Qingchun Liang , Wenwu Zeng , Yijun Peng , Zhenyu Zhao , Yiyi Liang , Jiadi Luo , Xiang Wang
show 1 more author
Shaoliang Peng
This is my paper · ORCID
classification eess.IVcs.AIcs.CV
keywords spreadthroughairspaceslungcancerwholeslideimagegraphneuralnetworkSiamesehistopathologymultipleinstancelearningcomputer-aideddiagnosis
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

Spread through air spaces (STAS) is an aggressive invasion pattern in lung cancer: tumor cells drift through air spaces beyond the main tumor, and its presence can change whether a surgeon removes only the nodule or a whole lobe. The paper proposes VERN, a feature-interactive Siamese graph encoder that predicts STAS directly from whole-slide histopathology images by turning each slide into a spatial graph of tissue patches. In their experiments VERN reached an AUC of 0.9215 in internal validation, 0.8275 on frozen test sections and 0.8829 on paraffin test sections, and held up across external validation cohorts. If these numbers hold, VERN would give pathologists and surgeons an automated second reader for a decision that currently hinges on subjective, time-consuming slide review.

What carries the argument

The load-bearing object is VERN itself, a feature-interactive Siamese graph encoder. Each whole-slide image is converted into a spatial topological graph: patches become nodes, patch features come from two pretrained extractors (1024- and 768-dimensional), and a K-nearest-neighbor rule (K=9) draws edges between nearby patches, so the graph encodes where tissue structures sit relative to one another. VERN then runs two symmetric encoder branches, each a sequence of GCNConv, SAGEConv, ReLU, Dropout, MLP and Rescale layers; the branches share weights and exchange information through cross-graph message passing, and a skip connection carries the original input forward. The two branch outputs are concatenated and averaged for the slide-level STAS prediction, and per-patch attention contributions are normalized and mapped back onto the slide as heatmaps.

What would settle it

Re-run the five-fold cross-validation with all slides grouped by patient before splitting, so no patient appears in both training and test folds, and compare the resulting AUC with the reported 0.9215; a large drop would show the model was exploiting patient-specific slide artifacts rather than STAS biology.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that STAS—a pattern in which tumor cells spread through alveolar air spaces beyond the main tumor—can be predicted from whole-slide histopathology images at the slide level, and that representing the slide as a spatial graph is what makes this work. The authors report an internal validation AUC of 0.9215, frozen-section and paraffin-section test AUCs of 0.8275 and 0.8829, and external cohort AUCs in the range of roughly 0.70 to 0.92. They also report that VERN outperforms five multiple instance learning baselines under the same five-fold cross-validation protocol. The paper frames this as the first whole-slide-image-level STAS prediction method and as a possible intraoperative aid, since frozen-section STAS is clinically important but hard for pathologists to detect.

Load-bearing premise

The load-bearing premise is that all slides from the same patient stayed in the same training or test split—if a patient's frozen and paraffin slides were allowed to straddle the split, the model could be recognizing patient-specific slide artifacts rather than STAS itself, which would inflate the reported AUCs.

Editorial extensions

If this is right

  • If VERN's accuracy holds, intraoperative frozen-section review gains a rapid second reader that could cut missed STAS cases before the surgeon chooses resection extent.
  • The per-patch attention maps give pathologists a concrete region to re-check, namely the tumor edge and peritumoral air spaces, rather than scanning the entire slide.
  • The graph representation is the claimed reason for the gains, so future whole-slide classifiers for other spatially distributed patterns could adopt the same encoder.
  • Because the model was trained on both frozen and paraffin sections, its frozen-section predictions benefit from paraffin-derived prior knowledge, a training strategy that could be reused elsewhere.

Reading between the lines

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

  • The paper does not report an ablation that replaces the K-nearest-neighbor spatial graph with random edges; such an ablation would directly test whether spatial topology, rather than the extra parameters alone, drives the improvement over multiple instance learning baselines.
  • Because each patient contributed multiple slides, a patient-level fusion rule (any slide positive implies STAS) and a patient-level AUC would make the clinical utility figures more interpretable; the paper mentions the rule but does not evaluate it.
  • The same architecture could be transferred to other histopathology tasks where spread or adjacency matters, such as lymphovascular invasion or perineural invasion, provided a similar frozen-plus-paraffin training set is available.
  • An external reader study comparing VERN against multiple pathologists on the same slides would place the reported AUCs in a clinical context; the paper compares with literature numbers rather than a head-to-head reader study.
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.

Referee Report

4 major / 4 minor

Summary. The paper proposes VERN, a feature-interactive Siamese graph encoder for predicting spread through air spaces (STAS) from lung cancer whole-slide images. The authors constructed a single-cohort dataset of 1,546 frozen and paraffin-embedded sections from 356 patients, trained VERN with five-fold cross-validation, and report an internal AUC of 0.9215 (best fold; average 0.8683), plus AUCs of 0.8275 and 0.8829 on frozen and paraffin test sections. They also validate on 356 single-cohort slides and on external slides from Zhengzhou/Henan Cancer Hospital, TCGA, and CPTAC, and provide an open web platform. The central claim is that VERN achieves clinical-grade STAS prediction and outperforms existing multiple-instance learning baselines owing to its spatial graph construction.

Significance. If the reported results hold, this would be one of the first WSI-level STAS predictors and a potentially useful clinical decision-support tool, especially given the low sensitivity of frozen-section STAS assessment by pathologists. The work has clear strengths: it compares against five MIL baselines, provides interpretability heatmaps, publishes code and an online platform, and includes three external datasets. However, the significance is conditional on the internal evaluation being unbiased; the manuscript currently does not establish patient-level independence in the cross-validation split, and the headline number is the best of five folds rather than a central estimate.

major comments (4)
  1. [Methods: Clinical single cohort and multicenter data collection; Results (first paragraph)] The internal cross-validation split is described only at the slide level: 'the internal validation set, which includes both FSs and PSs, into five subsets.' The manuscript never states that all slides from a single patient were kept in the same fold. Since 1,546 slides come from 356 patients, with each patient contributing one FS and several PSs, a slide-level split can place slides from the same patient in both training and validation. This would let the model exploit patient-specific staining, scanning, or tissue-preparation signatures rather than learning generalizable STAS morphology, directly inflating the reported average AUC of 0.8683 and the best-fold AUC of 0.9215, and also biasing the comparison against MIL baselines. Please specify whether patient-level grouping was used, or re-run the cross-validation with patient-stratified folds and report the resulting performance.
  2. [Results, Figure 2 and Table 1] The abstract and Results highlight an AUC of 0.9215, but this is the best of five cross-validation folds; the average in-domain AUC reported in Table 1 and Figure 2c is 0.8683. Selecting the best fold as the headline performance overstates the expected prospective accuracy. Please report all five fold-level AUCs with confidence intervals, and state explicitly whether any model selection was performed across folds. If the best fold was chosen post hoc, the reported performance is optimistic.
  3. [Results: Single-cohort and multicenter STAS validation sets; Methods: Clinical single cohort and multicenter data…] The 356-slide 'single-cohort validation' is not an independent patient-level validation set. The Methods state that these are additional PSs from the same single cohort and that they 'had not been internally trained, verified, and tested,' but the same 356 patients contributed FSs and PSs used in internal training. If any slide from a patient was in the training set, then the patient's STAS status and slide characteristics are known to the model during training, so the AUC of 0.9181 in Figure 4a reflects within-patient correlation and cannot be cited as evidence of generalizability. Please either restrict this validation to patients entirely excluded from training or reframe it as a within-cohort slide-type test, and rely on the truly external datasets for generalization claims.
  4. [Methods: Clinical single cohort and multicenter data collection] The external TCGA and CPTAC labels were assigned by two pathologists, but the manuscript reports no inter-observer agreement (e.g., Cohen's kappa), no description of the labeling protocol's blinding, and no verification against a reference standard. Given that STAS diagnosis is known to be subjective and that even intraoperative frozen-section assessment has moderate agreement, the reliability of these external labels is load-bearing for the external AUCs (0.7029 and 0.7555). Please report the labeling procedure in detail, including the criteria used, whether disagreements were adjudicated, and the inter-observer agreement.
minor comments (4)
  1. [Figure 2a caption] The caption reads 'with the diagonal line showing the in-domain test results'; in an ROC plot the diagonal line is chance-level performance, not the test results. Please clarify what the diagonal line represents.
  2. [Abstract and Results] The terms 'histopathological images,' 'slides,' and 'sections' are used interchangeably; please standardize terminology (e.g., 'whole-slide images' vs. 'patches') to avoid confusion about the evaluation unit.
  3. [Methods: Experimental setup and implementation details] The grid search is described only as tuning 'learning rate, regularization strength, and batch size,' but only the learning rate and optimizer settings are reported. Please list the ranges considered and the final hyperparameter values for reproducibility.
  4. [Results, Table 1] The table header contains 'PRC' in the caption text, but the table reports AUC and other metrics. Please ensure metric abbreviations are defined consistently (e.g., use AUROC and AUPRC).

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: VERN's STAS predictions are evaluated on held-out internal and external data, and no claim reduces to its inputs by construction.

full rationale

The paper is an empirical deep-learning study rather than a theoretical derivation chain. VERN is a GCN/SAGE-style graph encoder applied to KNN spatial graphs built from KimiaNet and CTransPath patch embeddings, and STAS predictions are measured on a held-out internal test set (40 FSs and 100 PSs), a single-cohort validation set of 356 PSs, two external hospital slide sets (91 slides), and TCGA/CPTAC external sets (101 and 100 slides). None of the model equations (Eq. 1, Eq. 2) defines the STAS label in terms of the model's own output, and no parameter is fitted to the test or external datasets and then relabeled as a prediction. Reporting the best cross-validation fold (AUC 0.9215) alongside the five-fold average (0.8683) is a reporting or selection concern, not a circularity; similarly, the absence of an explicit statement that all slides from a patient were kept in the same split is a potential leakage or optimism concern, but the paper does not exhibit an equation-level or definitional reduction of the target into the inputs. There is no load-bearing self-citation chain: the cited methods are standard public architectures (GCN, SAGE, MIL baselines) and externally validated benchmarks, and no uniqueness theorem or prior author result is invoked to force the choice of VERN. Because the central claim is supported by held-out and external evaluation and is not equivalent to its inputs by construction, the appropriate circularity score is 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities. Its scientific commitments are empirical: a dataset, a model architecture, and a set of hyperparameters. The free parameters listed are the hand-chosen values that materially affect the graph construction and training. The main unstated premises are the reliability of the STAS labels, the adequacy of the graph representation, and the validity of slide-level labels for a focal pathological pattern.

free parameters (5)
  • KNN neighborhood size K = 9
    Chosen for constructing the spatial topological graph from patch coordinates; no ablation or sensitivity analysis is reported, and the graph structure directly shapes the input to VERN.
  • Dropout rate = 0.2
    Set by hand in the VERN encoder; no reported sensitivity analysis.
  • Learning rate and RMSprop alpha = 0.001 and 0.9
    Tuned via grid search; these choices affect the final trained model and all reported metrics.
  • Training epochs = 200
    Chosen without reporting early stopping or convergence analysis.
  • Patch size and magnification = 512x512 pixels at 20x
    Segmentation choice for all WSIs; no ablation on patch size is provided.
assumptions (4)
  • domain assumption STAS status can be determined from H&E whole-slide images and the labels assigned by two pathologists are accurate ground truth.
    Invoked in Methods during data collection and labeling; no inter-observer agreement or independent label verification is reported, yet all training and evaluation depend on these labels.
  • domain assumption K-nearest neighbor graphs on patch coordinates capture the spatial relationship between the main tumor and STAS that is needed for classification.
    Stated in the Introduction and Discussion as the motivation for graph construction; no experiment isolates whether graph topology, rather than the features or the classifier, causes the reported improvement.
  • domain assumption A single slide-level label is valid even though STAS is a focal pattern, and averaging patch-level contributions yields a meaningful slide-level prediction.
    Implied by the MIL-style training and the prediction workflow; the paper does not analyze how many patches or how much tumor-adjacent tissue are needed for reliable STAS detection.
  • domain assumption Pretrained KimiaNet and CTransPath features transfer to this STAS prediction task.
    Both feature extractors were pretrained on other histopathology datasets; the paper assumes their representations contain the relevant morphology for STAS without fine-tuning on the target domain.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Feature-interactive Siamese graph encoder-based image analysis to predict STAS from histopathology images in lung cancer." pith.science (2026). https://pith.science/paper/IISQVLO5

@misc{pith2026241115274,
  author       = {Pith},
  title        = {Pith review of: Feature-interactive Siamese graph encoder-based image analysis to predict STAS from histopathology images in lung cancer},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IISQVLO5}},
  note         = {Machine review of arXiv:2411.15274}
}
read the original abstract

Spread through air spaces (STAS) is a distinct invasion pattern in lung cancer, crucial for prognosis assessment and guiding surgical decisions. Histopathology is the gold standard for STAS detection, yet traditional methods are subjective, time-consuming, and prone to misdiagnosis, limiting large-scale applications. We present VERN, an image analysis model utilizing a feature-interactive Siamese graph encoder to predict STAS from lung cancer histopathological images. VERN captures spatial topological features with feature sharing and skip connections to enhance model training. Using 1,546 histopathology slides, we built a large single-cohort STAS lung cancer dataset. VERN achieved an AUC of 0.9215 in internal validation and AUCs of 0.8275 and 0.8829 in frozen and paraffin-embedded test sections, respectively, demonstrating clinical-grade performance. Validated on a single-cohort and three external datasets, VERN showed robust predictive performance and generalizability, providing an open platform (http://plr.20210706.xyz:5000/) to enhance STAS diagnosis efficiency and accuracy.

Figures

Figures reproduced from arXiv: 2411.15274 by the authors.

Figure 1
Figure 1. Workflow including data preprocessing, model training and inference, interpretability analysis, and multicenter validation. a Data preprocessing involves digitization of WSI, extraction of regions of interest, segmentation of tissue into patches, and patch data augmentation. b The model architecture includes feature extraction from pretrained models (KimiaNet, CTransPath), construction of WSI spatial top ological ma… view at source ↗
Figure 2
Figure 2. Experimental results of STAS prediction using the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Interpretability analysis of the VERN. a WSIs with STAS-negative and STAS-positive annotations, where pathologists outlined the main tumor body and STAS subtypes, including micropapillary, solid nests, and single cells. b Attention scores for each patch based on the VERN. High values (red) indicate higher model-predicted contributions, while low values (purple) indicate lower contributions. Additionally, the nine pa… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Evaluation of VERN performance in predicting STAS in single-cohort and multicenter experiments. a ROC and PRC curves validating the VERN's effectiveness based on 356 histopathological images from the Second Xiangya Hospital of Central South University. b ROC and PRC cu…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

72 extracted references · 58 canonical work pages

  1. [1]

    Tumor islands in resected early -stage lung adenocarcinomas are associated with unique clinicopathologic and molecular characteristics and worse prognosis,

    M. L. Onozato et al., “Tumor islands in resected early -stage lung adenocarcinomas are associated with unique clinicopathologic and molecular characteristics and worse prognosis,” The American journal of surgical pathology, vol. 37, no. 2, pp. 287 –294, 2013

    work page 2013

  2. [2]

    Kadota et al

    K. Kadota et al. , “Tumor spread through air spaces is an important pattern of invasion and impacts the frequency and location of recurrences after limited resection for small stage I lung adenocarcinomas,” Journal of Thoracic Oncology, vol. 10, no. 5, pp. 806 –814, 2015

    work page 2015

  3. [3]

    Prognostic impact of intra -alveolar tumor spread in pulmonary adenocarcinoma,

    A. Warth et al. , “Prognostic impact of intra -alveolar tumor spread in pulmonary adenocarcinoma,” The American journal of surgical pathology , vol. 39, no. 6, pp. 793 –801, 2015

    work page 2015

  4. [4]

    Prognostic significance of tumor spread through air spaces in patients with stage IA part -solid lung adenocarcinoma after sublobar resection,

    M. Chae et al. , “Prognostic significance of tumor spread through air spaces in patients with stage IA part -solid lung adenocarcinoma after sublobar resection,” Lung Cancer, vol. 152, pp. 21–26, Feb. 2021, doi: 10.1016/j.lungcan.2020.12.001

    work page doi:10.1016/j.lungcan.2020.12.001 2021

  5. [5]

    Lobectomy is associated with better outcomes than sublobar resection in spread through air spaces (ST AS)- positive T1 lung adenocarcinoma: a propensity score–matched analysis,

    T. Eguchi et al., “Lobectomy is associated with better outcomes than sublobar resection in spread through air spaces (ST AS)- positive T1 lung adenocarcinoma: a propensity score–matched analysis,” Journal of thoracic oncology, vol. 14, no. 1, pp. 87– 98, 2019

    work page 2019

  6. [6]

    Spread through air spaces is a predictive factor of recurrence and a prognostic factor in stage I lung adenocarcinoma,

    S. Shiono and N. Yanagawa, “Spread through air spaces is a predictive factor of recurrence and a prognostic factor in stage I lung adenocarcinoma,” Interactive cardiovascular and thoracic surgery , vol. 23, no. 4, pp. 567 –572, 2016

    work page 2016

  7. [7]

    Tumor spread through air spaces affects the recurrence and overall survival in patients with lung adenocarcinoma> 2 to 3 cm,

    C. Dai et al. , “Tumor spread through air spaces affects the recurrence and overall survival in patients with lung adenocarcinoma> 2 to 3 cm,” Journal of thoracic oncology, vol. 12, no. 7, pp. 1052 –1060, 2017

    work page 2017

  8. [8]

    Pathological prognostic factors of recurrence in early stage lung adenocarcinoma,

    E. Yi, M. Bae, S. Cho, J. Chung, S. Jheon, and K. Kim, “Pathological prognostic factors of recurrence in early stage lung adenocarcinoma,” ANZ journal of surgery, vol. 88, no. 4, pp. 327 –331, 2018

    work page 2018

Show all 72 references
  1. [9]

    Spread through air spaces is an independent predictor of recurrence in stage III (N2) lung adenocarcinoma,

    Y. Terada et al., “Spread through air spaces is an independent predictor of recurrence in stage III (N2) lung adenocarcinoma,” Interactive cardiovascular and thoracic surgery, vol. 29, no. 3, pp. 442 –448, 2019

    2019

  2. [10]

    Analysis of recurrence in lung adenocarcinoma with spread through air spaces,

    H. A. Khalil et al. , “Analysis of recurrence in lung adenocarcinoma with spread through air spaces,” The Journal of Thoracic and Cardiovascular Surgery, vol. 166, no. 5, pp. 1317 –1328, 2023

    2023

  3. [11]

    STAS: New explorations and challenges for thoracic surgeons,

    T. Xia, Q. Yuan, and S. Xing, “STAS: New explorations and challenges for thoracic surgeons,” Clinical and Translational Oncology , pp. 1 –11, 2024

    2024

  4. [12]

    W . D. Travis et al., “The International Association for the Study of Lung Cancer (IASLC) Staging Project for Lung Cancer: Recommendation to Introduce Spread Through Air Spaces as a Histologic Descriptor in the Ninth Edition of the TNM Classification of Lung Cancer. Analysis o...

    2024 doi

  5. [13]

    Prognostic significance of limited resection in pathologic stage I lung adenocarcinoma with spread through air spaces,

    K. Xu et al., “Prognostic significance of limited resection in pathologic stage I lung adenocarcinoma with spread through air spaces,” Journal of Thoracic Disease, vol. 15, no. 9, p. 4795, 2023

    2023

  6. [14]

    Colorimetric histology using plasmonically active microscope slides,

    E. Balaur et al., “Colorimetric histology using plasmonically active microscope slides,” Nature, vol. 598, no. 7879, pp. 65– 71, Oct. 2021, doi: 10.1038/s41586 -021 -03835 -2

    2021 doi

  7. [15]

    Deep learning outperformed 11 pathologists in the classification of histopathological melanoma images,

    A. Hekler et al., “Deep learning outperformed 11 pathologists in the classification of histopathological melanoma images,” European Journal of Cancer, vol. 118, pp. 91–96, Sep. 2019, doi: 10.1016/j.ejca.2019.06.012

    2019 doi

  8. [16]

    Pathology Visions 2020: Through the Prism of Innovation,

    “Pathology Visions 2020: Through the Prism of Innovation,” Journal of Pathology Informatics , vol. 12, no. 1, p. 37, Jan. 2021, doi: 10.4103/2153 -3539.326643

    2020

  9. [17]

    Pathology: the clinical description of human disease,

    W . K. Funkhouser, “Pathology: the clinical description of human disease,” in Essential Concepts in Molecular Pathology, Elsevier, 2020, pp. 177 –190

    2020

  10. [18]

    The NanoSuit method: a novel histological approach for examining paraffin sections in a nondestructive manner by correlative light and electron microscopy ,

    H. Kawasaki et al., “The NanoSuit method: a novel histological approach for examining paraffin sections in a nondestructive manner by correlative light and electron microscopy ,” Laboratory Investigation, vol. 100, no. 1, pp. 161 –173, Jan. 2020, doi: 10.1038/s41374 -019 -0309 -7

    2020 doi

  11. [19]

    The value of frozen section diagnosis of tumor spread through air spaces in small -sized (≤ 2 cm) non -small cell lung cancer,

    Y. Ding et al., “The value of frozen section diagnosis of tumor spread through air spaces in small -sized (≤ 2 cm) non -small cell lung cancer,” World Journal of Surgical Oncology, vol. 21, no. 1, p. 195, Jul. 2023, doi: 10.1186/s12957-023 -03092 -9

    2023 doi

  12. [20]

    H. Cao et al., “Prediction of Spread Through Air Spaces (ST AS) By Intraoperative Frozen Section for Patients with cT1N0M0 Invasive Lung Adenocarcinoma: A Multi-Center Observational Study (ECTOP -1016),” Annals of Surgery , pp. 10 –1097

  13. [21]

    Accuracy and reproducibility of intraoperative assessment on tumor spread through air spaces in stage 1 lung adenocarcinomas,

    J. A. Villalba et al. , “Accuracy and reproducibility of intraoperative assessment on tumor spread through air spaces in stage 1 lung adenocarcinomas,” Journal of Thoracic Oncology , vol. 16, no. 4, pp. 619 –629, 2021

    2021

  14. [22]

    Assessment of the feasibility of frozen sections for the detection of spread through air spaces (STAS) in pulmonary adenocarcinoma,

    F. Zhou et al., “Assessment of the feasibility of frozen sections for the detection of spread through air spaces (STAS) in pulmonary adenocarcinoma,” Modern Pathology, vol. 35, no. 2, pp. 210 –217, 2022

    2022

  15. [23]

    Artificial intelligence in radiology ,

    A. Hosny, C. Parmar, J. Quackenbush, L. H. Schwartz, and H. J. W . L. Aerts, “Artificial intelligence in radiology ,” N at Rev Cancer, vol. 18, no. 8, pp. 500 –510, Aug. 2018, doi: 10.1038/s41568 -018 -0016 -5

    2018 doi

  16. [24]

    Recent advances and applications of deep learning methods in materials science,

    K. Choudhary et al., “Recent advances and applications of deep learning methods in materials science,” npj Computational Materials, vol. 8, no. 1, p. 59, Apr. 2022, doi: 10.1038/s41524- 022 -00734 -6

    2022 doi

  17. [25]

    Semi - supervised training of deep convolutional neural networks with heterogeneous data and few local annotations: An experiment on prostate histopathology image classification,

    N. Marini, S. Otálora, H. Müller, and M. Atzori, “Semi - supervised training of deep convolutional neural networks with heterogeneous data and few local annotations: An experiment on prostate histopathology image classification,” Medical Image Analysis , vol. 73, p. 102165, Oc...

    2021

  18. [26]

    Classification and mutation prediction based on histopathology H&E images in liver cancer using deep learning,

    M. Chen et al., “Classification and mutation prediction based on histopathology H&E images in liver cancer using deep learning,” npj Precis. Onc., vol. 4, no. 1, p. 14, Dec. 2020, doi: 10.1038/s41698 -020 -0120 -3

    2020 doi

  19. [27]

    MIST: multiple instance learning network based on Swin Transformer for whole slide image classification of colorectal adenomas,

    H. Cai et al., “MIST: multiple instance learning network based on Swin Transformer for whole slide image classification of colorectal adenomas,” The Journal of Pathology, vol. 259, no. 2, pp. 125 –135, 2023

    2023

  20. [28]

    Deconv -transformer (DecT): A histopathological image classification model for breast cancer based on color deconvolution and transformer architecture,

    Z. He et al., “Deconv -transformer (DecT): A histopathological image classification model for breast cancer based on color deconvolution and transformer architecture,” Information Sciences, vol. 608, pp. 1093 –1112, Aug. 2022, doi: 12 10.1016/j.ins.2022.06.091

    2022 doi

  21. [29]

    Topological Tumor Graphs: A Graph - Based Spatial Model to Infer Stromal Recruitment for Immunosuppression in Melanoma Histology,

    H. Failmezger, S. Muralidhar, A. Rullan, C. E. de Andrea, E. Sahai, and Y. Yuan, “Topological Tumor Graphs: A Graph - Based Spatial Model to Infer Stromal Recruitment for Immunosuppression in Melanoma Histology,” Cancer Research, vol. 80, no. 5, pp. 1199 –1209, Mar. 2020, doi:...

    2020 doi

  22. [31]

    Integrating spatial gene expression and breast tumour morphology via deep learning,

    B. He et al. , “Integrating spatial gene expression and breast tumour morphology via deep learning,” Nature Biomedical Engineering , vol. 4, no. 8, pp. 827 –834, Aug. 2020, doi: 10.1038/s41551 -020 -0578 -x

    2020 doi

  23. [32]

    Attention -based deep multiple instance learning,

    M. Ilse, J. Tomczak, and M. Welling, “Attention -based deep multiple instance learning,” presented at the International conference on machine learning, PMLR, 2018, pp. 2127–2136

    2018

  24. [33]

    Dual-stream multiple instance learning network for whole slide image classification with self- supervised contrastive learning,

    B. Li, Y. Li, and K. W . Eliceiri, “Dual-stream multiple instance learning network for whole slide image classification with self- supervised contrastive learning,” presented at the Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 14...

    2021

  25. [34]

    Transmil: Transformer based correlated multiple instance learning for whole slide image classification,

    Z. Shao, H. Bian, Y. Chen, Y. Wang, J. Zhang, and X. Ji, “Transmil: Transformer based correlated multiple instance learning for whole slide image classification,” Advances in Neural Information Processing Systems , vol. 34, pp. 2136 – 2147, 2021

    2021

  26. [35]

    DTFD-MIL: Double-Tier Feature Distillation Multiple Instance Learning for Histopathology Whole Slide Image Classification,

    H. Zhang et al., “DTFD-MIL: Double-Tier Feature Distillation Multiple Instance Learning for Histopathology Whole Slide Image Classification,” in 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , New Orleans, LA, USA: IEEE, Jun. 2022, pp. 18780 –18790...

    2022

  27. [36]

    Interventional Bag Multi -Instance Learning On Whole -Slide Pathological Images,

    T. Lin, Z. Yu, H. Hu, Y. Xu, and C. W . Chen, “Interventional Bag Multi -Instance Learning On Whole -Slide Pathological Images,” in 2023 IEEE/CVF Conference on Computer V ision and Pattern Recognition (CVPR), 2023, pp. 19830–19839. doi: 10.1109/CVPR52729.2023.01899

    2023

  28. [37]

    Nuclear morphology is a deep learning biomarker of cellular senescence,

    I. Heckenbach et al., “Nuclear morphology is a deep learning biomarker of cellular senescence,” Nature Aging, vol. 2, no. 8, pp. 742 –755, Aug. 2022, doi: 10.1038/s43587-022-00263 -3

    2022 doi

  29. [38]

    A comprehensive overview and comparative analysis on deep learning models: CNN, RNN, LSTM, GRU,

    F. M. Shiri, T. Perumal, N. Mustapha, and R. Mohamed, “A comprehensive overview and comparative analysis on deep learning models: CNN, RNN, LSTM, GRU,” arXiv preprint arXiv:2305.17473 , 2023

    2023 arXiv

  30. [39]

    Gross specimen handling procedures do not impact the occurrence of spread through air spaces (STAS) in lung cancer,

    J. Metovic et al., “Gross specimen handling procedures do not impact the occurrence of spread through air spaces (STAS) in lung cancer,” The American journal of surgical pathology, vol. 45, no. 2, pp. 215 –222, 2021

    2021

  31. [40]

    Histopathological aspects of resected non -small cell lung cancer, with emphasis on spread through air spaces and collapsed adenocarcinoma in situ,

    J. L. G. Blaauwgeers, “Histopathological aspects of resected non -small cell lung cancer, with emphasis on spread through air spaces and collapsed adenocarcinoma in situ,” 2023

    2023

  32. [41]

    Graph learning considering dynamic structure and random structure,

    H. Dong, H. Ma, Z. Du, Z. Zhou, H. Yang, and Z. W ang, “Graph learning considering dynamic structure and random structure,” Journal of King Saud University -Computer and Information Sciences , vol. 35, no. 7, p. 101633, 2023

    2023

  33. [42]

    Gas exchange and ventilation– perfusion relationships in the lung,

    J. Petersson and R. W . Glenny, “Gas exchange and ventilation– perfusion relationships in the lung,” European Respiratory Journal , vol. 44, no. 4, pp. 1023 –1041, 2014

    2014

  34. [43]

    Global variations in lung cancer incidence by histological subtype in 2020: a population -based study ,

    Y. Zhang et al., “Global variations in lung cancer incidence by histological subtype in 2020: a population -based study ,” The Lancet Oncology , vol. 24, no. 11, pp. 1206 –1218, 2023

    2020

  35. [44]

    Prognostic impact of tumor spread through air spaces for T2aN0 stage IB non‐small cell lung cancer,

    Z. Chen et al., “Prognostic impact of tumor spread through air spaces for T2aN0 stage IB non‐small cell lung cancer,” C ancer Medicine, vol. 12, no. 14, pp. 15246 –15255, 2023

    2023

  36. [45]

    Correlation of tumor spread through air spaces and clinicopathological characteristics in surgically resected lung adenocarcinomas,

    S.-Y. Hu et al., “Correlation of tumor spread through air spaces and clinicopathological characteristics in surgically resected lung adenocarcinomas,” Lung Cancer, vol. 126, pp. 189–193, 2018

    2018

  37. [46]

    Relationship between MTA1 and spread through air space and their joint influence on prognosis of patients with stage I-III lung adenocarcinoma,

    Y. Liu et al., “Relationship between MTA1 and spread through air space and their joint influence on prognosis of patients with stage I-III lung adenocarcinoma,” Lung cancer, vol. 124, pp. 211–218, 2018

    2018

  38. [47]

    Genetic and clinicopathologic characteristics of lung adenocarcinoma with tumor spread through air spaces,

    J. S. Lee, E. K. Kim, M. Kim, and H. S. Shim, “Genetic and clinicopathologic characteristics of lung adenocarcinoma with tumor spread through air spaces,” Lung Cancer, vol. 123, pp. 121 –126, 2018

    2018

  39. [48]

    p53 immunohistochemistry is an accurate surrogate for TP53 mutational analysis in endometrial carcinoma biopsies,

    N. Singh et al. , “p53 immunohistochemistry is an accurate surrogate for TP53 mutational analysis in endometrial carcinoma biopsies,” The Journal of Pathology, vol. 250, no. 3, pp. 336 –345, 2020

    2020

  40. [49]

    Correlative assessment of p53 immunostaining patterns and TP53 mutation status by next - generation sequencing in high-grade endometrial carcinomas,

    N. Matsumoto et al. , “Correlative assessment of p53 immunostaining patterns and TP53 mutation status by next - generation sequencing in high-grade endometrial carcinomas,” International Journal of Gynecological Pathology, vol. 42, no. 6, pp. 567 –575, 2023

    2023

  41. [50]

    Concordance of ALK fusion gene - rearrangement between immunohistochemistry and next - generation sequencing,

    K. Wakuda et al. , “Concordance of ALK fusion gene - rearrangement between immunohistochemistry and next - generation sequencing,” International Journal of Clinical Oncology , vol. 29, no. 2, pp. 96 –102, 2024

    2024

  42. [51]

    Integration of clinicopathological and mutational data offers insight into lung cancer with tumor spread through air spaces,

    Y. Tian et al. , “Integration of clinicopathological and mutational data offers insight into lung cancer with tumor spread through air spaces,” Annals of Translational Medicine, vol. 9, no. 12, 2021

    2021

  43. [52]

    An individualized nomogram for predicting and validating spread through air space (STAS) in surgically resected lung adenocarcinoma: a single center retrospective analysis,

    J. Wang, Y. Yao, D. Tang, and W . Gao, “An individualized nomogram for predicting and validating spread through air space (STAS) in surgically resected lung adenocarcinoma: a single center retrospective analysis,” Journal of Cardiothoracic Surgery , vol. 18, no. 1, p. 337, 2023

    2023

  44. [53]

    Comprehensive analysis of spread through air spaces in lung adenocarcinoma and squamous cell carcinoma using the 8th edition AJCC/UICC staging system,

    M. Jia, S. Yu, J. Yu, Y. Li, H. Gao, and P. -L. Sun, “Comprehensive analysis of spread through air spaces in lung adenocarcinoma and squamous cell carcinoma using the 8th edition AJCC/UICC staging system,” BMC cancer, vol. 20, pp. 1–11, 2020

    2020

  45. [54]

    Spread through air spaces (STAS) is a predictor of poor outcome in atypical carcinoids of the lung,

    S. Altinay et al. , “Spread through air spaces (STAS) is a predictor of poor outcome in atypical carcinoids of the lung,” Virchows Archiv, vol. 475, pp. 325 –334, 2019

    2019

  46. [55]

    Dynamic monitoring of PD‐L1 and Ki67 in circulating tumor cells of metastatic non‐small cell lung cancer patients treated with pembrolizumab,

    M. Spiliotaki et al., “Dynamic monitoring of PD‐L1 and Ki67 in circulating tumor cells of metastatic non‐small cell lung cancer patients treated with pembrolizumab,” Molecular Oncology , vol. 17, no. 5, pp. 792 –809, 2023

    2023

  47. [56]

    Spectral imaging with deep learning,

    L. Huang, R. Luo, X. Liu, and X. Hao, “Spectral imaging with deep learning,” Light: Science & Applications, vol. 1 1, no. 1, p. 61, 2022

    2022

  48. [57]

    Fine-Tuning and training of densenet for histopathology image representation using TCGA diagnostic slides,

    A. Riasatian et al., “Fine-Tuning and training of densenet for histopathology image representation using TCGA diagnostic slides,” Medical Image Analysis, vol. 70, p. 102032, 2021, doi: https://doi.org/10.1016/j.media.2021.102032

    2021

  49. [58]

    Transformer-based unsupervised contrastive learning for histopathological image classification,

    X. Wang et al., “Transformer-based unsupervised contrastive learning for histopathological image classification,” Medical Image Analysis , vol. 81, p. 102559, Oct. 2022, doi: 10.1016/j.media.2022.102559

    2022

  50. [59]

    Densely Connected Convolutional Networks,

    G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinber ger , “Densely Connected Convolutional Networks,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI: IEEE, Jul. 2017, pp. 2261–2269. doi: 13 10.1109/CVPR.2017.243

    2017 doi

  51. [60]

    Swin transformer: Hierarchical vision transformer using shifted windows,

    Z. Liu et al. , “Swin transformer: Hierarchical vision transformer using shifted windows,” presented at the Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 10012 –10022

    2021

  52. [61]

    A fuzzy k-nearest neighbor algorithm,

    J. M. Keller, M. R. Gray, and J. A. Givens, “A fuzzy k-nearest neighbor algorithm,” IEEE transactions on systems, man, and cybernetics, no. 4, pp. 580 –585, 1985

    1985

  53. [62]

    Pathomic Fusion: An Integrated Framework for Fusing Histopathology and Genomic Features for C ancer Diagnosis and Prognosis,

    R. J. Chen et al., “Pathomic Fusion: An Integrated Framework for Fusing Histopathology and Genomic Features for C ancer Diagnosis and Prognosis,” IEEE Transactions on Medical Imaging , vol. 41, no. 4, pp. 757 –770, Apr. 2022, doi: 10.1109/TMI.2020.3021387

    2022

  54. [63]

    Semi -supervised classification with graph convolutional networks,

    T. N. Kipf and M. Welling, “Semi -supervised classification with graph convolutional networks,” arXiv preprint arXiv:1609.02907 , 2016

    2016 arXiv

  55. [64]

    Inductive Representation Learning on Large Graphs,

    W . L. Hamilton, R. Ying, and J. Leskovec, “Inductive Representation Learning on Large Graphs,” in Proceedings of the 31st International Conference on Neural Information Processing Systems, in NIPS’17. Red Hook, NY, USA: C urran Associates Inc., 2017, pp. 1025 –1035

    2017

  56. [65]

    Deep Learning using Rectified Linear Units (ReLU),

    A. F. Agarap, “Deep Learning using Rectified Linear Units (ReLU),” arXiv:1803.08375 [cs, stat], Feb. 2019, Accessed: Mar. 15, 2020. [Online]. Available: http://arxiv.org/abs/1803.08375

    2019 arXiv

  57. [66]

    Local-aggregation graph networks,

    J. Chang, L. Wang, G. Meng, Q. Zhang, S. Xiang, and C . Pan, “Local-aggregation graph networks,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 42, no. 1 1, pp. 2874 –2886, 2019

    2019

  58. [67]

    Multilayer perceptron (MLP),

    H. Taud and J. -F. Mas, “Multilayer perceptron (MLP),” Geomatic approaches for modeling land change scenarios, pp. 451 –455, 2018

    2018

  59. [68]

    Prevalence of neural collapse during the terminal phase of deep learning training,

    V. Papyan, X. Han, and D. L. Donoho, “Prevalence of neural collapse during the terminal phase of deep learning training,” Proceedings of the National Academy of Sciences, vol. 1 17, no. 40, pp. 24652 –24663, 2020

    2020

  60. [69]

    Uctransnet: rethinking the skip connections in u -net from a channel-wise perspective with transformer,

    H. Wang, P. Cao, J. Wang, and O. R. Zaiane, “Uctransnet: rethinking the skip connections in u -net from a channel-wise perspective with transformer,” presented at the Proceedings of the AAAI conference on artificial intelligence, 2022, pp. 2441– 2449

    2022

  61. [70]

    Deep learning modelling techniques: current progress, applications, advantages, and challenges,

    S. F. Ahmed et al. , “Deep learning modelling techniques: current progress, applications, advantages, and challenges,” Artificial Intelligence Review, vol. 56, no. 11, pp. 13521–13617, 2023

    2023

  62. [71]

    A sufficient condition for convergences of adam and rmsprop,

    F. Zou, L. Shen, Z. Jie, W . Zhang, and W . Liu, “A sufficient condition for convergences of adam and rmsprop,” presented at the Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition, 2019, pp. 11127 –11135

    2019

  63. [72]

    The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation,

    D. Chicco and G. Jurman, “The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation,” BMC genomics, vol. 21, pp. 1–13, 2020

    2020

  64. [73]

    Comparison of deep learning approaches to predict COVID-19 infection,

    T. B. Alakus and I. Turkoglu, “Comparison of deep learning approaches to predict COVID-19 infection,” Chaos, Solitons & Fractals, vol. 140, p. 110120, 2020

    2020

Pith tools

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