Pith. sign in

REVIEW 3 major objections 5 minor 34 references

Identifying Signatures of Image Phenotypes to Track Treatment Response in Liver Disease

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A deep clustering network learns a vocabulary of liver MRI tissue patterns that separates NASH treatment arms and predicts biopsy-derived histology, outperforming established non-invasive measures.

desk verdict A useful proof-of-concept for unsupervised MRI tissue vocabularies in NASH, but the headline claim of better treatment-group separation than HFF/ALT is not actually tested. read the letter →

arxiv 2507.12012 v1 pith:N4WXP2QV submitted 2025-07-16 eess.IV cs.AIcs.CV

Matthias Perkonigg , Nina Bastati , Ahmed Ba-Ssalamah , Peter Mesenbrink , Alexander Goehler , Miljen Martic , Xiaofei Zhou , Michael Trauner
show 1 more author
Georg Langs
This is my paper · ORCID
classification eess.IVcs.AIcs.CV
keywords deepclusteringtissuevocabularyliverMRINASHtreatmentresponsebiomarkerdiscoveryunsupervisedlearningimagesignatures
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that an unsupervised neural network can learn a vocabulary of recurring MRI tissue patterns in the liver, and that the relative frequency of these patterns—a per-patient 'signature'—quantifies disease progression and treatment response in non-alcoholic steatohepatitis (NASH). Using data from a randomized placebo-controlled trial, the authors show that changes in these signatures separate high-dose treatment groups from placebo and low-dose groups better than established endpoints such as hepatic fat fraction and ALT. The same signatures predict biopsy-derived histological grades for steatosis, ballooning, and inflammation, and support the discovery of patient phenotypes associated with disease markers. If correct, the approach offers a non-invasive, repeatable way to monitor diffuse liver disease and could serve as a quantitative endpoint in drug development.

What carries the argument

The core mechanism is a Deep Clustering Network (DCN), an autoencoder whose latent space is jointly optimized with a k-means clustering objective. Patches from multi-parametric liver MRI (T1-weighted, Dixon, and six-echo sequences) are encoded into a 20-dimensional latent space and assigned to one of K clusters; the composition of a liver is summarized by a signature vector counting relative cluster frequencies. For longitudinal data, signature differences between visits are fed to a random forest regressor to predict treatment group, while registered cluster maps yield a transition matrix $M_{ij}$ describing the probability that tissue class $i$ at baseline becomes class $j$ at follow-up. Supervised random forest classifiers and regressors map signatures to histology grades, and hierarchical agglomerative clustering over signatures defines patient phenotypes.

What would settle it

Train a DCN using only patches from training-fold patients, with held-out patients completely excluded from the unsupervised step, and repeat the 5-fold treatment-separation and biopsy-prediction experiments; if the separation collapses, the original performance depended on test-patient information rather than a generalizable tissue vocabulary. Alternatively, train a classifier to predict the MRI scanner or site from a patient's signature vector; if it succeeds with high accuracy, the vocabulary may be encoding site-specific artifacts rather than disease biology.

Watch

Extended reading notes

Core claim

The central claim is that quantifiable image phenotypes—recurring tissue-appearance patterns learned without supervision—carry clinically meaningful information about liver disease that is not captured by standard non-invasive measurements. The paper demonstrates this on a randomized controlled trial of NASH patients: a signature formed from the relative frequencies of five learned clusters in each of three MRI sequences, fused across sequences, separates patients receiving 140 mcg or 200 mcg of the study drug from those receiving placebo or low dose, with statistically significant differences (e.g., placebo vs. 200 mcg, t=4.710, p=0.0001). The same signature predicts biopsy grades for ballooning and inflammation on both the trial cohort and a separate clinical-routine replication cohort, and tissue transition maps obtained by registering baseline and follow-up scans reveal treatment-specific pathways, such as transitions from steatosis-correlated to steatosis-anticorrelated tissue components.

Load-bearing premise

The vocabulary learned from MRI patches is a stable, site-independent measure of liver tissue, despite the multi-center trial data involving 43 scanners and no explicit cross-site normalization.

Editorial extensions

If this is right

  • Repeated MRI during therapy could replace some repeated biopsies for monitoring NASH patients.
  • The significant separation of the 200 mcg group from placebo indicates the signatures may be sensitive enough to serve as a quantitative endpoint in early-phase drug trials.
  • Registered transition maps identify which specific tissue classes change under treatment, linking imaging response to histology-related features like steatosis.
  • Because the vocabulary is learned unsupervised, it could be reused across different liver diseases or imaging protocols without manual labeling.
  • The comparable results on a separate single-scanner clinical cohort suggest the method can transfer from trial settings to routine clinical practice.

Reading between the lines

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

  • Inference: The main untested confound is scanner and site identity, since the DCN was trained on patches from all patients, including those later held out; a test of whether site can be predicted from the signatures would clarify whether the reported separation reflects biology or scanner-specific appearance.
  • Inference: The vocabulary approach could plausibly transfer to other diffuse-organ diseases such as kidney or lung fibrosis, but the current evidence is limited to liver MRI and no claim about other organs is made.
  • Inference: The biological grounding of individual clusters is only indirect (univariate correlation with histology); co-registering signatures with quantitative imaging or spatially matched histology would test whether each vocabulary element corresponds to a distinct tissue alteration.
  • Inference: A per-patient signature of fixed cluster count could be explored as a stratification tool in trial design, for example to identify non-responders early, though the paper does not evaluate such an application.
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 / 5 minor

Summary. This paper proposes an unsupervised approach to learn a "tissue vocabulary" from multi-parametric liver MRI patches using deep clustering networks (DCN). The vocabulary is used to build per-patient signatures (cluster frequency histograms) and to perform four analyses on a randomized controlled trial of tropifexor in NASH: (i) random forest regression from visit-to-visit signature differences to treatment dose, compared with changes in HFF and ALT; (ii) identification of tissue transition paths between baseline and follow-up; (iii) prediction of biopsy-based histology grades, with a replication on a clinical routine cohort; and (iv) discovery of imaging phenotypes. The abstract claims that the method enables better separation between treatment groups than established non-imaging measures and that the vocabulary can predict biopsy-derived features.

Significance. If the central claim were fully supported, this would be a useful contribution: the method produces interpretable, spatially localized tissue patterns from standard MRI, and the authors provide an ablation study, a code repository, and a separate replication cohort for histology prediction. The transition analysis is a nice step beyond global biomarkers. However, the headline claim of superiority over HFF/ALT is not backed by any statistical comparison, and the unsupervised representation is learned on all patients before the patient-level cross-validation, so the current evidence is weaker than the abstract suggests. With additional analysis, the work could be a solid methodological contribution.

major comments (3)
  1. [Section 4.3, Figure 2, Table 2] The statement in the abstract and Section 4.3 that the signatures "enable a better separation between treatment groups than established non-imaging measures" is not supported by the reported statistics. Table 2 contains only t-tests comparing SF-5-3 predictions between dose groups; no test compares the discriminative performance of SF-5-3 against HFF or ALT, and Figure 2 shows only unlabeled density curves. The authors should add a formal comparison, for instance AUC or C-index for the high-dose vs. low-dose/placebo contrast computed from SF-5-3 change, HFF change, and ALT change, with confidence intervals and a paired test (e.g., DeLong or bootstrap). Without such a comparison, the superiority claim is unsubstantiated.
  2. [Section 3.1.2, 4.3, 4.5] The DCN is trained on patches sampled from all patients, including the patients who are later in the test folds of the 5-fold random forest evaluation. Because the vocabulary is therefore influenced by the test patients' images, and because Section 5 acknowledges that no cross-site normalization was performed across the 43 scanners in the trial, the reported separation and histology prediction may partly reflect scanner or site structure rather than a transferable tissue phenotype. The authors should assess this by nested cross-validation (training the DCN within each fold) or by a site/scanner-stratified analysis. The CR-pred replication does not resolve the issue, because its sequences differ from the RCT and the DCN training data for that cohort are not specified.
  3. [Section 4.3] The grouping of "placebo and Low dose vs. high doses of 140mcg and 200mcg" is used to support the main conclusion, but it is introduced after inspecting the data and is not tested as a specific contrast. The pairwise t-tests in Table 2 do not establish that this particular two-group separation is significant after accounting for the post hoc choice of grouping. The authors should pre-specify or formally test the contrast, and report an effect size and confidence interval for the difference.
minor comments (5)
  1. [Abstract and Section 4.3] HFF is an MRI-based measure, so calling HFF and ALT "non-imaging measures" in the abstract is inaccurate; "established endpoints" would be clearer.
  2. [Section 4.5] For the CR-pred replication, the text should state explicitly whether the DCN was retrained on CR-pred images or transferred from the RCT; this matters for interpreting the replication.
  3. [Section 4.3] The phrase "random forest regressor to predict the treatment arm" is unusual because treatment arm is categorical; please clarify how the regression target was encoded (e.g., numeric dose) and why regression was chosen over classification.
  4. [Figure 2] The x-axis is described as "mirrored HFF and ALT," but the direction of mirroring is not explained; please clarify so the comparison is interpretable.
  5. [Table 1] The header "Treat. Group Treat. Arm" appears to be a formatting artifact; also the table would benefit from a clear statement of which patients overlap between RCT-pred and RCT-progress.

Circularity Check

2 steps flagged · score 4.0 of 10

The label-based derivation is not circular, but the held-out evaluations leak test patients into the unsupervised vocabulary and the SF-5-3 cluster setting is selected on the same evaluation data, undermining the claim that the results are predictions from a prespecified model.

  1. other [Figure 1 caption and Section 4.3 (also Section 4.5)]
    "To train the DCN, image patches are sampled randomly from the liver of all patients and the DCN is trained. The trained encoder of the network is applied to each patient by extracting and encoding a patch to a cluster at each position within the liver, the relative frequency of clusters is used as image signature. ... Due to the limited size of the data set, 5-fold cross validation was used to separate training- and test set."

    The DCN that defines the tissue vocabulary is trained on patches from every patient, with no patient-level split before training. The 5-fold cross-validation described in Sections 4.3 and 4.5 is applied only to the downstream random forest, not to the DCN. Therefore, in each fold, the held-out test patient's signature is computed by assigning that patient's patches to cluster centroids that were fitted using those same patches (along with all others). The 'held-out' evaluation is thus transductive: the representation has already seen the test patient's raw images. This can encode patient- or scanner-specific appearance into the vocabulary and inflate treatment-arm separation and biopsy-prediction accuracy.

  2. fitted input called prediction [Section 4.2 and Section 4.7.2]
    "In the following, unless stated otherwise, we use SF-5-3 signatures for all experiments, as this setting has proven to be most universal in preliminary experiments."

    The paper does not report a separate validation set for this hyperparameter choice. Section 4.7.2 compares SF-5-3, SF-8-3, and SF-10-3 on RCT-pred (Table 5), and then Section 4.5 reports RCT-pred biopsy-prediction results for the selected SF-5-3 configuration. Because the cluster-number selection is made on the same evaluation data before the 5-fold cross-validation, the reported accuracy is an optimistic selected result rather than an independent evaluation of a prespecified model. The treatment-response experiment in Section 4.3 also inherits this selected configuration. This is a milder form of fitting a hyperparameter to the test data and then calling the outcome a prediction.

full rationale

The two marker-prediction pipelines are not circular in the label sense: histology grades and treatment-arm labels enter only in the supervised random forest step, and the random forest itself is evaluated in held-out folds. The external replication cohort (CR-pred) provides some independent support for biopsy-prediction transfer. However, two evaluation choices weaken the claim that these are predictions from a model fixed before seeing the evaluation patients. First, the DCN vocabulary is trained on patches from all patients, including those later placed in held-out folds, so the 'test' signatures are generated by a representation that has already seen those patients' images; this is transductive leakage rather than label circularity. Second, the SF-5-3 cluster count is selected by comparing cluster counts on RCT-pred, and the same data are then used to report the selected model, which inflates accuracy. The abstract's claim that the signatures enable 'a better separation between treatment groups than established non-imaging measures' is not supported by any statistical comparison of SF-5-3 against HFF/ALT: Figure 2 shows density curves only and Table 2 reports t-tests only for SF-5-3, with no paired test or test statistic for HFF/ALT. That is an evidentiary gap rather than a circularity, and it is noted separately. No load-bearing self-citation chain was found; the method relies on the external DCN reference [11] and random forests [26]. On balance, the central derivation is not forced by labels or by self-citation, so the circularity score is moderate.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The central claims rest on the assumption that MRI patch appearance clusters are tissue phenotypes and that the same vocabulary transfers across patients, scanners, and sequences. The paper uses several hand-chosen hyperparameters, and the phenotype groups are derived by clustering the same signatures used for evaluation.

free parameters (4)
  • Number of clusters per sequence K = 5 (SF-5-3)
    Hyperparameter chosen after ablations on RCT-pred (Section 4.7.2); the signature dimension and all downstream results depend on K.
  • Trade-off parameter lambda = not reported
    Introduced in Eq. 3 to balance reconstruction and clustering losses; no value is given in the paper, yet it shapes the learned vocabulary.
  • Latent space dimension = 20
    Architecture choice in Section 3.1.2; controls the compression of image patches and therefore the clustering behavior.
  • Number of phenotypes P = 7 (A-G)
    Threshold of the hierarchical clustering in Section 4.6; the reported odds ratios depend on this grouping choice.
assumptions (4)
  • domain assumption Image patches sampled inside the liver mask reflect underlying tissue biology rather than scanner artifacts.
    Central premise for all signature-based claims; the paper notes scanner variability is not normalized beyond basic image normalization (Section 5).
  • domain assumption The deep clustering vocabulary trained on all patients is valid for new patients and transfers across scanners and sequences.
    No patient-level split is described before DCN training, and CR-pred uses different MRI sequences than the RCT, so transferability is assumed (Sections 4.2 and 4.5).
  • domain assumption 3D rigid registration aligns baseline and 12-week follow-up livers well enough for voxel-level cluster transitions to represent tissue change.
    Section 3.2.2 uses 3D rigid registration on a deformable abdominal organ; misregistration could create spurious transitions.
  • domain assumption Histopathology grades and the NAFLD fibrosis score are reliable enough as ground truth.
    The paper itself cites sampling variability of biopsy [3,4] and notes noisy labels in Section 5, yet uses these grades as the supervised targets.
invented entities (1)
  • Tissue vocabulary components (cluster types)
    purpose: To form patient-level signatures and tissue transition maps from MRI patches.
    The clusters are data-derived groupings treated as tissue types, but no independent biological or pathological validation is provided beyond correlations with biopsy grades in the same cohorts.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Identifying Signatures of Image Phenotypes to Track Treatment Response in Liver Disease." pith.science (2026). https://pith.science/paper/N4WXP2QV

@misc{pith2026250712012,
  author       = {Pith},
  title        = {Pith review of: Identifying Signatures of Image Phenotypes to Track Treatment Response in Liver Disease},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N4WXP2QV}},
  note         = {Machine review of arXiv:2507.12012}
}
read the original abstract

Quantifiable image patterns associated with disease progression and treatment response are critical tools for guiding individual treatment, and for developing novel therapies. Here, we show that unsupervised machine learning can identify a pattern vocabulary of liver tissue in magnetic resonance images that quantifies treatment response in diffuse liver disease. Deep clustering networks simultaneously encode and cluster patches of medical images into a low-dimensional latent space to establish a tissue vocabulary. The resulting tissue types capture differential tissue change and its location in the liver associated with treatment response. We demonstrate the utility of the vocabulary on a randomized controlled trial cohort of non-alcoholic steatohepatitis patients. First, we use the vocabulary to compare longitudinal liver change in a placebo and a treatment cohort. Results show that the method identifies specific liver tissue change pathways associated with treatment, and enables a better separation between treatment groups than established non-imaging measures. Moreover, we show that the vocabulary can predict biopsy derived features from non-invasive imaging data. We validate the method on a separate replication cohort to demonstrate the applicability of the proposed method.

Figures

Figures reproduced from arXiv: 2507.12012 by the authors.

Figure 1
Figure 1. Overview of the proposed method. To train the DCN, image patches are sampled [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Quantitative separation of treatment arms based on MRI signatures compared to estab [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Transitions of liver tissue between tissue classes from baseline to follow-up, evaluated [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Random forest regression based prediction of biopsy grades of ballooning, inflammation, [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: Identifying phenotypes and testing their association with histo-pathology markers. Clus [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Benefit of multi-squence assessment: Comparison of using different sequences on [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Random forest regression based prediction of biopsy grades of ballooning, in￾flammation, and steatosis, and NALFD fi￾brosis score with SF-5-3. T-test with Bon￾ferroni multiple comparison correction was applied: * 0.01 < p <= 0.05, ** 0.05 < p <= 0.001, *** 0.001 < p <=…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 31 canonical work pages

  1. [1]

    Current Status in Testing for Nonalcoholic Fatty Liver Disease (NAFLD) and Nonalcoholic Steatohepatitis (NASH),

    H. Drescher, S. Weiskirchen, and R. Weiskirchen, “Current Status in Testing for Nonalcoholic Fatty Liver Disease (NAFLD) and Nonalcoholic Steatohepatitis (NASH),” Cells, vol. 8, no. 8, p. 845, 8 2019

    work page 2019

  2. [2]

    Diagnostic and treatment implications of nonalcoholic fatty liver disease and nonalcoholic steatohepatitis,

    T. Wong, R. J. Wong, and R. G. Gish, “Diagnostic and treatment implications of nonalcoholic fatty liver disease and nonalcoholic steatohepatitis,” Gastroenterology and Hepatology, vol. 15, no. 2, pp. 83–89, 2 2019

    work page 2019

  3. [3]

    Sampling variability of liver biopsy in nonalcoholic fatty liver disease,

    V. Ratziu, F. Charlotte, A. Heurtier, S. Gombert, P. Giral, E. Bruckert, A. Grimaldi, F. Capron, and T. Poynard, “Sampling variability of liver biopsy in nonalcoholic fatty liver disease,” Gastroenterology, vol. 128, no. 7, pp. 1898–1906, 2005

    work page 1906

  4. [4]

    Effects of Liver Biopsy Sample Length and Number of Readings on Sampling Variability in Nonalcoholic Fatty Liver Disease,

    R. Vuppalanchi, A. ¨Unalp, M. L. Van Natta, O. W. Cummings, K. E. Sandrasegaran, T. Hameed, J. Tonascia, and N. Chalasani, “Effects of Liver Biopsy Sample Length and Number of Readings on Sampling Variability in Nonalcoholic Fatty Liver Disease,” Clinical Gastroenterology and Hepatology, vol. 7, no. 4, pp. 481–486, 4 2009

    work page 2009

  5. [5]

    Noninvasive biomarkers in NAFLD and NASH — current progress and future promise,

    V. Wai-Sun Wong, L. A. Adams, V. de L´ edinghen, G. Lai-Hung Wong, and S. Sookoian, “Noninvasive biomarkers in NAFLD and NASH — current progress and future promise,” Nature Reviews Gastroenterology & Hepatology , 2018. [Online]. Available: https://doi.org/10.1038/

    work page 2018

  6. [6]

    Deep learning enables pathologist-like scoring of NASH models,

    F. Heinemann, G. Birk, and B. Stierstorfer, “Deep learning enables pathologist-like scoring of NASH models,” Scientific Reports, vol. 9, no. 1, pp. 1–10, 12 2019

    work page 2019

  7. [7]

    Training of deep convolutional neural networks to identify critical liver alterations in histopathology image samples,

    A. Arjmand, C. T. Angelis, V. Christou, A. T. Tzallas, M. G. Tsipouras, E. Glavas, R. Forlano, P. Manousou, and N. Giannakeas, “Training of deep convolutional neural networks to identify critical liver alterations in histopathology image samples,” Applied Sciences (Switzerland) , vol. 10, no. 1, 2020. 15

    work page 2020

  8. [8]

    Application of Machine Learn- ing Methods to Predict Non-Alcoholic Steatohepatitis (NASH) in Non-Alcoholic Fatty Liver (NAFL) Patients,

    S. Fialoke, A. Malarstig, M. R. Miller, and A. Dumitriu, “Application of Machine Learn- ing Methods to Predict Non-Alcoholic Steatohepatitis (NASH) in Non-Alcoholic Fatty Liver (NAFL) Patients,” AMIA ... Annual Symposium proceedings. AMIA Symposium , vol. 2018, pp. 430–439, 2018

    work page 2018

Show all 34 references
  1. [9]

    Development and validation of a deep learning system for staging liver fibrosis by using contrast agent–enhanced CT images in the liver,

    K. J. Choi, J. K. Jang, S. S. Lee, Y. S. Sung, W. H. Shim, H. S. Kim, J. Yun, J. Y. Choi, Y. Lee, B. K. Kang, J. H. Kim, S. Y. Kim, and E. S. Yu, “Development and validation of a deep learning system for staging liver fibrosis by using contrast agent–enhanced CT images in the ...

    2018

  2. [10]

    Correlation of histologic, imaging, and artificial intelligence features in nafld patients, derived from gd-eob-dtpa-enhanced mri: a proof-of- concept study,

    N. Bastati, M. Perkonigg, D. Sobotka, S. Poetter-Lang, R. Fragner, A. Beer, A. Messner, M. Watzenboeck, S. Pochepnia, J. Kittinger, A. Herold, A. Kristic, J. C. Hodge, S. Traussnig, M. Trauner, A. Ba-Ssalamah, and G. Langs, “Correlation of histologic, imaging, and artificial i...

    2023

  3. [11]

    Towards K-means-friendly Spaces: Simultaneous Deep Learning and Clustering,

    B. Yang, X. Fu, N. D. Sidiropoulos, and M. Hong, “Towards K-means-friendly Spaces: Simultaneous Deep Learning and Clustering,” in Proceedings of the 34th International Conference on Machine Learning , 2017. [Online]. Available: https: //github.com/boyangumn/DCN

    2017

  4. [12]

    Unsupervised Deep Embedding for Clustering Analysis,

    J. Xie, R. Girshick, and A. Farhadi, “Unsupervised Deep Embedding for Clustering Analysis,” in Proceedings of the 33 rd International Conference on Machine Learning, vol. 48,

  5. [13]

    Discriminatively Boosted Image Clustering with Fully Convolutional Auto-Encoders,

    F. Li, H. Qiao, B. Zhang, and X. Xi, “Discriminatively Boosted Image Clustering with Fully Convolutional Auto-Encoders,” Pattern Recognition, vol. 83, pp. 161–173, 3 2017. [Online]. Available: http://arxiv.org/abs/1703.07980

    2017 arXiv

  6. [14]

    Deep Clustering via Joint Convolutional Autoencoder Embedding and Relative Entropy Minimization,

    K. G. Dizaji, A. Herandi, C. Deng, W. Cai, and H. Huang, “Deep Clustering via Joint Convolutional Autoencoder Embedding and Relative Entropy Minimization,” Proceedings of the IEEE International Conference on Computer Vision , vol. 2017-October, pp. 5747–5756, 4

    2017

  7. [15]

    Variational Deep Embedding: An Unsupervised and Generative Approach to Clustering,

    Z. Jiang, Y. Zheng, H. Tan, B. Tang, and H. Zhou, “Variational Deep Embedding: An Unsupervised and Generative Approach to Clustering,” in Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (IJCAI-17) , 11 2017. [Online]. Available: http:/...

    2017 arXiv

  8. [16]

    Joint Unsupervised Learning of Deep Representations and Image Clusters,

    J. Yang, D. Parikh, D. Batra, and V. Tech, “Joint Unsupervised Learning of Deep Representations and Image Clusters,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016. [Online]. Available: https://github.com/

    2016

  9. [17]

    Deep Adaptive Image Clustering,

    J. Chang, L. Wang, G. Meng, S. Xiang, and C. Pan, “Deep Adaptive Image Clustering,” in Proceedings of the IEEE International Conference on Computer Vision , vol. 2017-October. Institute of Electrical and Electronics Engineers Inc., 12 2017, pp. 5880–5888

    2017

  10. [18]

    Learning Discrete Representations via Information Maximizing Self-Augmented Training,

    W. Hu, T. Miyato, S. Tokui, E. Matsumoto, and M. Sugiyama, “Learning Discrete Representations via Information Maximizing Self-Augmented Training,” 34th International Conference on Machine Learning, ICML 2017 , vol. 4, pp. 2467–2481, 2 2017. [Online]. Available: http://arxiv.or...

    2017 arXiv

  11. [19]

    Unsupervised Segmentation of 3D Medical Images Based on Clustering and Deep Representation Learning,

    T. Moriya, H. R. Roth, S. Nakamura, H. Oda, K. Nagara, M. Oda, and K. Mori, “Unsupervised Segmentation of 3D Medical Images Based on Clustering and Deep Representation Learning,” Medical Imaging 2018: Biomedical Applications in Molecular, Structural, and Functional Imaging , v...

    2018 arXiv

  12. [20]

    Semi-supervised learning with deep embed- ded clustering for image classification and segmentation,

    J. Enguehard, P. O’Halloran, and A. Gholipour, “Semi-supervised learning with deep embed- ded clustering for image classification and segmentation,” IEEE Access, vol. 7, pp. 11 093– 11 104, 2019

    2019

  13. [21]

    Alzheimer’s disease diagnosis based on multiple cluster dense convolutional networks,

    F. Li and M. Liu, “Alzheimer’s disease diagnosis based on multiple cluster dense convolutional networks,” Computerized Medical Imaging and Graphics , vol. 70, pp. 101–110, 12 2018. 16

    2018

  14. [22]

    Unsupervised Joint Mining of Deep Features and Image Labels for Large-scale Radiology Image Categorization and Scene Recognition,

    X. Wang, L. Lu, H.-c. Shin, L. Kim, M. Bagheri, I. Nogues, J. Yao, and R. M. Summers, “Unsupervised Joint Mining of Deep Features and Image Labels for Large-scale Radiology Image Categorization and Scene Recognition,” in IEEE Winter Conference on Applications of Computer Visio...

    2017

  15. [23]

    Data heterogeneity and continual machine learning for medical image anal- ysis,

    M. Perkonigg, “Data heterogeneity and continual machine learning for medical image anal- ysis,” PhD thesis, Medical University of Vienna, Vienna, Austria, 2022, available at https: //repositorium.meduniwien.ac.at/obvumwhs/content/titleinfo/7786301

    2022

  16. [24]

    Least Squares Quantization in PCM,

    S. P. Lloyd, “Least Squares Quantization in PCM,” Tech. Rep. 2, 1982

    1982

  17. [25]

    Video Google: A Text Retrieval Approach to Object Matching in Videos,

    J. Sivic and A. Zisserman, “Video Google: A Text Retrieval Approach to Object Matching in Videos,” in ICCV ’03: Proceedings of the Ninth IEEE International Conference on Computer Vision, 2003

    2003

  18. [26]

    Random Forests,

    L. Breiman, “Random Forests,” Tech. Rep., 2001

    2001

  19. [27]

    Modern hierarchical, agglomerative clustering algorithms,

    D. M¨ ullner, “Modern hierarchical, agglomerative clustering algorithms,” no. 1973, pp. 1–29,

    1973

  20. [28]

    Lucas, P

    K. Lucas, P. Lopez, E. Lawitz, A. Sheikh, D. Aizenberg, S. Hsia, G. B. Bee, J. Vierling, J. Frias, J. White, Y. Eguchi, D. Lazas, G. Neff, M. Yoneda, S. Augustin, W. Kim, Y. Loeffler, F. Schaefer, S. Lamle, M. Martic, C. Brass, and A. Sanyal, “Tropifexor, a highly potent FXR a...

    2020

  21. [29]

    The NAFLD fibrosis score: A noninvasive system that identifies liver fibrosis in patients with NAFLD,

    P. Angulo, J. M. Hui, G. Marchesini, E. Bugianesi, J. George, G. C. Farrell, F. Enders, S. Saksena, A. D. Burt, J. P. Bida, K. Lindor, S. O. Sanderson, M. Lenzi, L. A. Adams, J. Kench, T. M. Therneau, and C. P. Day, “The NAFLD fibrosis score: A noninvasive system that identifi...

    2007

  22. [30]

    PyTorch: An Imperative Style, High-Performance Deep Learning Library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury Google, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. K. Xamla, E. Yang, Z. Devito, M. Raison Nabla, A. Tejani, S. Chilamkurthy, Q. Ai, B. Steiner, L. F. Facebook, J. B. Facebook, and S. Chintala,...

    2019

  23. [31]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015, N. Navab, J. Hornegger, W. M. Wells, and A. F. Frangi, Eds. Cham: Springer International Pub...

    2015

  24. [2011]

    Available: http://arxiv.org/abs/1109.2378

    [Online]. Available: http://arxiv.org/abs/1109.2378

    arXiv

  25. [2016]

    Available: http://arxiv.org/abs/1511.06335

    [Online]. Available: http://arxiv.org/abs/1511.06335

    arXiv

  26. [2017]

    Available: http://arxiv.org/abs/1704.06327

    [Online]. Available: http://arxiv.org/abs/1704.06327

    arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.