Pith. sign in

REVIEW 4 major objections 9 minor 61 references

4D VQ-GAN: Synthesising Medical Scans at Any Time Point for Personalised Disease Progression Modelling of Idiopathic Pulmonary Fibrosis

T0 review · 4 major / 9 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read Given two CT scans of an IPF patient, 4D-VQ-GAN synthesises realistic CT volumes at any requested time point, and biomarkers from those synthetic scans predict survival nearly as well as biomarkers from real scans.

desk verdict The model combination is new and the survival check is a real external validation, but the paper has not yet shown that its generated scans beat a no-change baseline, so the progression claims need work before publication. read the letter →

arxiv 2502.05713 v1 pith:K3JYY5IF submitted 2025-02-08 eess.IV cs.AIcs.CVcs.LG

An Zhao , Moucheng Xu , Ahmed H. Shahin , Wim Wuyts , Mark G. Jones , Joseph Jacob , Daniel C. Alexander This is my paper · ORCID
classification eess.IVcs.AIcs.CVcs.LG
keywords 4DimagesynthesisVQ-GANneuralordinarydifferentialequationsdiseaseprogressionmodellingCTimagingidiopathicpulmonaryfibrosissurvivalanalysis
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

Idiopathic pulmonary fibrosis is a progressive lung disease whose future course is hard to predict, and CT scans that could show that course are rarely taken because they involve radiation. This paper tries to establish that a two-stage generative model, 4D-VQ-GAN, can take two CT scans of the same IPF patient at different times and synthesise realistic CT volumes at any requested time point, producing an individualised trajectory of disease progression. The authors argue the generated scans are clinically meaningful: imaging biomarkers extracted from synthetic follow-up scans predict survival with a C-index of 0.886 for cross-sectional markers and 0.959 for one-year longitudinal changes, close to 0.929 and 1.0 obtained from real scans. If the claim holds, it would give clinicians a way to visualise probable disease evolution from sparse data and plan treatment earlier.

What carries the argument

The central object is the 4D-VQ-GAN, a two-stage architecture that couples a 3D-VQ-GAN with a latent neural ODE. The 3D-VQ-GAN learns a discrete codebook of volumetric image patterns and reconstructs CT volumes from quantised embeddings; the neural ODE $\mathrm{d}h/\mathrm{d}t = f_\theta(h,t)$ then defines how those embeddings drift over time. A 3D-ConvGRU encoder compresses the observed embedding sequence into an initial state $h_{t_0}$, the ODE solver produces hidden states at arbitrary target times, and a convolutional projector converts the difference between successive hidden states into a difference map added to the previous embedding. This difference-plus-skip-connection design is what lets the model generate a continuous trajectory from only two observations.

What would settle it

A decisive test would be to take two scans from patients with documented different IPF subtypes or rates of decline and compare the model's generated third scans; if the synthetic future lungs are statistically indistinguishable across groups, the patient-specific progression claim fails. Quantitatively, the claim would be falsified if, on a held-out cohort stratified by subtype, the C-index of biomarkers from generated scans drops to chance (0.5) once subtype is accounted for.

Watch

Extended reading notes

Core claim

The central claim is that continuous disease progression in volumetric CT can be learned in a discrete latent space rather than modelled directly on voxels. A 3D-VQ-GAN first compresses each CT volume into a compact set of codebook indices whose entries correspond to local imaging patterns such as reticulation and ground-glass opacity. A second-stage temporal model, built around a neural ODE $f_\theta$, evolves the latent embedding over time; given two observed scans, the ODE solver is run from the encoded initial state to produce embeddings at any queried timestep, and skip connections between consecutive embeddings let the model learn only the difference between adjacent time points. The frozen 3D-VQ-GAN decoder then turns those embeddings back into CT volumes. The paper demonstrates interpolation and extrapolation on IPF patients and validates the generated scans through Cox survival analysis, reporting that biomarkers from generated scans achieve a C-index of 0.886 (cross-sectional) and 0.959 (longitudinal change) versus 0.929 and 1.0 for real scans.

Load-bearing premise

One fixed mathematical rule is assumed to describe how every patient's lungs change over time, and the paper itself says this is unlikely for a disease with distinct subtypes and progression patterns.

Editorial extensions

If this is right

  • Given two CT scans, the model outputs synthetic scans at any requested interval and horizon, covering both interpolation and extrapolation.
  • The best extrapolation configuration (ConvGRU encoder with embedding skip connections) reaches MSE 0.019, SSIM 0.489, and PSNR 17.816 on held-out patients.
  • Survival-relevant texture is preserved: cross-sectional biomarkers from generated third scans give C-index 0.886 versus 0.929 for real scans, and longitudinal biomarkers give 0.959 versus 1.0.
  • Synthetic longitudinal scans could be used to impute missing visits, augment sparse IPF datasets, and simulate treatment effects by comparing generated trajectories under different assumptions.

Reading between the lines

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

  • Beyond the paper, the same two-stage recipe could transfer to other diseases with sparse longitudinal imaging, such as multiple sclerosis or osteoarthritis, because it only requires a learnable discrete codebook and a smooth latent trajectory.
  • The paper's shared disease dynamic assumption is the obvious pressure point; conditioning the ODE on a per-patient latent code or subtype label would be a natural, untested upgrade that makes the personalised claim more literal.
  • A sharp testable extension would be to feed generated scans into a separate segmentation or staging model and measure whether performance improves; if the synthetic volumes carry the same prognostic texture as real ones, they could function as data augmentation in rare-disease cohorts.
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 / 9 minor

Summary. This manuscript proposes 4D-VQ-GAN, a two-stage generative model for synthesizing longitudinal CT volumes in IPF patients. Stage 1 trains a 3D-VQ-GAN to compress CT volumes into a discrete codebook; Stage 2 trains a neural ODE on the quantized latent embeddings, conditioned on two observed scans, to generate volumes at arbitrary time points. The authors evaluate interpolation and extrapolation quality with MSE/SSIM/PSNR, perform an ablation over architectural variants, and validate clinical utility by deriving imaging biomarkers from generated scans and computing C-indices for survival prediction. They report C-indices of 0.886 (cross-sectional) and 0.959 (longitudinal) for generated scans, compared with 0.929 and 1.0 for real scans. The paper is clearly written, but the experimental validation is not sufficient to establish the central claim that the model captures disease progression in a personalized manner.

Significance. The paper addresses a relevant and difficult problem: generating longitudinal 3D medical images from sparse, irregular observations. The two-stage design (VQ-GAN compression + latent ODE) is sensible, and the use of a real longitudinal IPF cohort with survival outcomes is a strength. The codebook visualization and the external validation of the reconstruction stage are also commendable. However, as presented, the experimental validation does not establish the central claim that the model 'captures disease progression' or provides 'personalised' predictions. The absence of trivial baselines, the lack of statistical comparison for survival C-indices, and the untested assumption of a shared dynamics mean the quantitative results could be explained by simpler mechanisms such as copying the last observed scan or reflecting baseline severity. If the required baselines and analyses confirm the added value, the contribution would be significant for the medical imaging community.

major comments (4)
  1. [Section 4, Table 1] The ablation compares only variants of the proposed architecture (ConvGRU vs ODE-ConvGRU, skip connections, masked inputs), with no comparison against trivial baselines such as copying the last observed scan or its latent embedding, or linear interpolation in latent space. Since IPF changes between scans can be subtle, the reported extrapolation SSIM of 0.489 and PSNR of 17.816 may be close to a no-change baseline; without such comparisons, the quantitative results do not demonstrate that the model predicts progression beyond the information already present in the input scans.
  2. [Section 4, Survival outcome prediction] The claim that generated scans yield C-indices 'comparable' to real scans is not supported by appropriate statistical evidence: no confidence intervals, no significance tests, and the test set is small (44 patients). The longitudinal C-index of 1.0 for real scans suggests overfitting or an overly homogeneous test set, and the cross-sectional difference (0.886 vs 0.929) is not shown to be within sampling variability. Moreover, there is no comparison with biomarkers derived from the most recent real scan, so the high C-index of generated scans may reflect baseline disease severity rather than the model's predicted temporal change.
  3. [Section 2, Eq. (6); Section 5] The model uses a single deterministic neural ODE f_theta, which the authors acknowledge 'assumes a shared disease dynamic across all patients.' This assumption is not tested, yet the title and introduction claim 'personalised disease progression modelling.' To support this claim, the authors should demonstrate that generated trajectories vary meaningfully across patients (e.g., by testing whether patient-specific initial states lead to distinct predicted changes, or by comparing against a model with patient-specific dynamics).
  4. [Section 3, Evaluation Metrics] The quantitative evaluation relies on image-similarity metrics (MSE, SSIM, PSNR) that do not directly measure whether the generated changes correspond to pathological progression. The authors should include clinically meaningful metrics, such as changes in quantitative fibrosis extent or the imaging biomarkers already used in their survival analysis, to show that the generated temporal changes are in the right direction and of plausible magnitude.
minor comments (9)
  1. [Table 1] The variant names are confusing; for example, 'ODE encoder (ConvGRU)' and 'ODE encoder (ODE-ConvGRU)' could be more clearly distinguished, and the table should state which components are ablated in each row.
  2. [Section 4] The paper should report 95% confidence intervals for all C-indices (e.g., via bootstrap) and a test of the difference between generated and real biomarkers.
  3. [Appendix D.2, Eq. (3)] The GAN loss is written for a single discriminator, but the text mentions two discriminators D2d and D3d; the equation should include both terms explicitly.
  4. [Sections 2 and D.3] The 3D-ConvGRU is described as the 'encoder of the neural ODE solver', but it is actually the encoder that produces the initial latent state; rephrase to avoid confusion.
  5. [Section 3, Datasets] Provide more details on the test cohort: number of patients, number of scans per patient, and the distribution of time intervals between scans.
  6. [Appendix A] The phrase 'visual inspection by me' should be rewritten in a professional style, e.g., 'visual inspection by a trained researcher.'
  7. [Figure 3] Add annotations indicating which generated volumes correspond to which time points and which are interpolated versus extrapolated.
  8. [Appendix D.4, Table 3] The reconstruction errors on the external Southampton test set are considerably higher than on the internal test set; this deserves a comment in the main text.
  9. [Related Work] The paper would benefit from a comparison with at least one existing longitudinal image synthesis method, such as the latent linear mixed-effects model of Sauty et al. or TR-GAN, to contextualize the results.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the two-stage latent ODE model is trained on real CT volumes and validated against held-out survival outcomes, with no fitted parameter renamed as a prediction.

full rationale

The derivation chain is self-contained and externally validated. Stage 1 learns a discrete codebook by reconstructing real CT volumes; Stage 2 fits a neural ODE to the temporal trajectory of those codes; neither stage defines its target in terms of the evaluation. The survival analysis uses real, held-out survival outcomes: the top five code-frequency biomarkers are selected on training scans by Cox p-values, and the C-index is then computed on a separate test set for both real and generated scans, so no fitted parameter is renamed as a prediction. The only couplings—the same codebook is used for generation and biomarker extraction, and the deterministic ODE assumes a shared disease dynamic—are potential biases or modelling limitations, not reductions of the claimed result to its inputs. The reported absence of a 'no-change' baseline is a validation gap that affects how strongly the progression claim is supported, but it does not make the derivation circular. No self-citation chain or imported uniqueness theorem is load-bearing.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claim rests on the assumptions listed. The main free parameters are architectural choices (codebook size, compression rate) calibrated on internal/external reconstruction error, and the biomarker count in the survival pipeline. No invented physical entities are introduced; the 'codebook imaging patterns' are learned features, not postulates.

free parameters (4)
  • Codebook vocabulary size M = 256
    Chosen by validation on reconstruction MSE on internal and external test sets (Table 3); affects the granularity of latent representation.
  • Compression rate r = 4
    Chosen as best balance between reconstruction quality and computational cost (Table 3).
  • Loss weights for 3D-VQ-GAN = lambda_perc=4, lambda_rec=4, lambda_GAN=1
    Adopted from prior work [16]; calibrated for reconstruction quality, affect the learned codebook.
  • Number of top biomarkers selected = 5
    The number of biomarkers used in survival analysis was fixed at five, selected by p-values from a Cox model on training data; this choice affects the reported C-index.
assumptions (3)
  • domain assumption The latent embedding space of the 3D-VQ-GAN supports smooth, continuous disease trajectories that can be described by a single neural ODE fθ shared across all patients.
    Eq. 6 defines the ODE; Section 5 acknowledges that a deterministic shared ODE may not capture heterogeneous disease subtypes.
  • domain assumption Normalized frequencies of codebook indices serve as meaningful prognostic imaging biomarkers.
    Section 2 (survival analysis) assumes the histogram of code indices reflects fibrosis patterns that correlate with survival; no validation of this assumption is provided beyond the Cox model results.
  • domain assumption Lung registration correctly aligns longitudinal scans so that residual differences reflect disease progression, not motion or positioning.
    Appendix A describes registration and visual inspection by the author, acknowledging that verification by radiologists was not performed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of 4D VQ-GAN: Synthesising Medical Scans at Any Time Point for Personalised Disease Progression Modelling of Idiopathic Pulmonary Fibrosis." pith.science (2026). https://pith.science/paper/K3JYY5IF

@misc{pith2026250205713,
  author       = {Pith},
  title        = {Pith review of: 4D VQ-GAN: Synthesising Medical Scans at Any Time Point for Personalised Disease Progression Modelling of Idiopathic Pulmonary Fibrosis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K3JYY5IF}},
  note         = {Machine review of arXiv:2502.05713}
}
read the original abstract

Understanding the progression trajectories of diseases is crucial for early diagnosis and effective treatment planning. This is especially vital for life-threatening conditions such as Idiopathic Pulmonary Fibrosis (IPF), a chronic, progressive lung disease with a prognosis comparable to many cancers. Computed tomography (CT) imaging has been established as a reliable diagnostic tool for IPF. Accurately predicting future CT scans of early-stage IPF patients can aid in developing better treatment strategies, thereby improving survival outcomes. In this paper, we propose 4D Vector Quantised Generative Adversarial Networks (4D-VQ-GAN), a model capable of generating realistic CT volumes of IPF patients at any time point. The model is trained using a two-stage approach. In the first stage, a 3D-VQ-GAN is trained to reconstruct CT volumes. In the second stage, a Neural Ordinary Differential Equation (ODE) based temporal model is trained to capture the temporal dynamics of the quantised embeddings generated by the encoder in the first stage. We evaluate different configurations of our model for generating longitudinal CT scans and compare the results against ground truth data, both quantitatively and qualitatively. For validation, we conduct survival analysis using imaging biomarkers derived from generated CT scans and achieve a C-index comparable to that of biomarkers derived from the real CT scans. The survival analysis results demonstrate the potential clinical utility inherent to generated longitudinal CT scans, showing that they can reliably predict survival outcomes.

Figures

Figures reproduced from arXiv: 2502.05713 by the authors.

Figure 1
Figure 1. The overview of our two-stage training strategies. The first stage trains an encoder-decoder-based 3D-VQ [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The inference of the trained model. Given two scans, the model can generate more scans up to the specified [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Three real CT scans of an IPF patient are shown in the upper panel, representing axial, coronal, and sagittal [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Segmentation results for selected cases from Leuven cohort [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Visualization of four registration outcomes with a focus on lung areas for clarity. The left two columns present [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Examples of input 3D CT scans and corresponding CT scans reconstructed by 3D-VQ-GAN with varying [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: The overview of the proposed two-stage model. The left side is the 3D-VQ-GAN for image reconstruction. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: The overview of the 3D-VQ-GAN for image reconstruction. [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: A visualization of the codebook (M = 256) Every entry in the codebook corresponds to a distinctive representation or code assigned to a specific region or pattern within the input space. Utilizing the techniques outlined in [23], we visually represent each code in the …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

61 extracted references · 50 canonical work pages

  1. [1]

    Delving deeper into convolutional networks for learning video representations

    Nicolas Ballas et al. “Delving deeper into convolutional networks for learning video representations”. In: arXiv preprint arXiv:1511.06432 (2015)

    arXiv 2015

  2. [2]

    Seeing what a GAN cannot generate

    David Bau et al. “Seeing what a GAN cannot generate”. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019, pp. 4502–4511

    work page 2019

  3. [3]

    Estimating or propagating gradients through stochastic neurons for conditional computation

    Yoshua Bengio, Nicholas Léonard, and Aaron Courville. “Estimating or propagating gradients through stochastic neurons for conditional computation”. In: arXiv preprint arXiv:1308.3432 (2013)

    arXiv 2013

  4. [4]

    Neural ordinary differential equations

    Ricky TQ Chen et al. “Neural ordinary differential equations”. In: Advances in neural information processing systems. V ol. 31. 2018

    work page 2018

  5. [5]

    Longitudinal self-supervision to disentangle inter- patient variability from disease progression

    Raphaël Couronné, Paul Vernhet, and Stanley Durrleman. “Longitudinal self-supervision to disentangle inter- patient variability from disease progression”. In: International Conference on Medical image computing and computer-assisted intervention. Springer. 2021, pp. 231–241

    work page 2021

  6. [6]

    Regression models and life-tables

    David R Cox. “Regression models and life-tables”. In: Journal of the Royal Statistical Society: Series B (Methodological) 34.2 (1972), pp. 187–202

    work page 1972

  7. [7]

    Generative adversarial networks: An overview

    Antonia Creswell et al. “Generative adversarial networks: An overview”. In:IEEE signal processing magazine 35.1 (2018), pp. 53–65

    work page 2018

  8. [8]

    Diffusion models in vision: A survey

    Florinel-Alin Croitoru et al. “Diffusion models in vision: A survey”. In: IEEE Transactions on Pattern Analysis and Machine Intelligence 45.9 (2023), pp. 10850–10869

    work page 2023

Show all 61 references
  1. [9]

    Estimating long-term multivariate progression from short-term data

    Michael C Donohue et al. “Estimating long-term multivariate progression from short-term data”. In: Alzheimer’s & Dementia 10 (2014), S400–S410. 6 A PREPRINT - F EBRUARY 11, 2025

    2014

  2. [10]

    Taming transformers for high-resolution image synthesis

    Patrick Esser, Robin Rombach, and Bjorn Ommer. “Taming transformers for high-resolution image synthesis”. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021, pp. 12873–12883

    2021

  3. [11]

    TR-Gan: multi-session future MRI prediction with temporal recurrent generative adversar- ial Network

    Chen-Chen Fan et al. “TR-Gan: multi-session future MRI prediction with temporal recurrent generative adversar- ial Network”. In: IEEE Transactions on Medical Imaging 41.8 (2022), pp. 1925–1937

    2022

  4. [12]

    TR-TransGAN: Temporal Recurrent Transformer Generative Adversarial Network for Longitudinal MRI Dataset Expansion

    Chen-Chen Fan et al. “TR-TransGAN: Temporal Recurrent Transformer Generative Adversarial Network for Longitudinal MRI Dataset Expansion”. In: IEEE Transactions on Cognitive and Developmental Systems (2024), pp. 1–10

    2024

  5. [13]

    GAN-based generation of realistic 3D data: A systematic review and taxonomy

    André Ferreira et al. “GAN-based generation of realistic 3D data: A systematic review and taxonomy”. In: arXiv preprint arXiv:2207.01390 (2022)

    2022 arXiv

  6. [14]

    GAN-based generation of realistic 3D volumetric data: A systematic review and taxonomy

    André Ferreira et al. “GAN-based generation of realistic 3D volumetric data: A systematic review and taxonomy”. In: Medical Image Analysis (2024), p. 103100

    2024

  7. [15]

    A fast operator for detection and precise location of distinct points, corners and centres of circular features

    Wolfgang Förstner and Eberhard Gülch. “A fast operator for detection and precise location of distinct points, corners and centres of circular features”. In: Proc. ISPRS intercommission conference on fast processing of photogrammetric data. V ol. 6. 1987, pp. 281–305

    1987

  8. [16]

    Long video generation with time-agnostic VQGAN and time-sensitive transformer

    Songwei Ge et al. “Long video generation with time-agnostic VQGAN and time-sensitive transformer”. In: European Conference on Computer Vision. Springer. 2022, pp. 102–118

    2022

  9. [17]

    GraphRegNet: Deep graph regularisation networks on sparse keypoints for dense registration of 3D lung CTs

    Lasse Hansen and Mattias P Heinrich. “GraphRegNet: Deep graph regularisation networks on sparse keypoints for dense registration of 3D lung CTs”. In: IEEE Transactions on Medical Imaging 40.9 (2021), pp. 2246–2257

    2021

  10. [18]

    Estimating large lung motion in COPD patients by symmetric regularised correspondence fields

    Mattias P Heinrich, Heinz Handels, and Ivor JA Simpson. “Estimating large lung motion in COPD patients by symmetric regularised correspondence fields”. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer. 2015, pp. 338–345

    2015

  11. [19]

    Learn2Reg: Comprehensive multi-task medical image registration challenge, dataset and evaluation in the era of deep learning

    Alessa Hering et al. “Learn2Reg: Comprehensive multi-task medical image registration challenge, dataset and evaluation in the era of deep learning”. In: IEEE Transactions on Medical Imaging 42.3 (2022), pp. 697–712

    2022

  12. [20]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. “Denoising diffusion probabilistic models”. In: Advances in Neural Information Processing Systems. V ol. 33. 2020, pp. 6840–6851

    2020

  13. [21]

    Imagen video: High definition video generation with diffusion models

    Jonathan Ho et al. “Imagen video: High definition video generation with diffusion models”. In: arXiv preprint arXiv:2210.02303 (2022)

    2022 arXiv

  14. [22]

    Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem

    Johannes Hofmanninger et al. “Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem”. In: European Radiology Experimental 4.1 (2020), pp. 1–13

    2020

  15. [23]

    Topological Neural Discrete Representation Learning à la Kohonen

    Kazuki Irie, Róbert Csordás, and Jürgen Schmidhuber. “Topological Neural Discrete Representation Learning à la Kohonen”. In: arXiv preprint arXiv:2302.07950 (2023)

    2023 arXiv

  16. [24]

    Brain β-amyloid load approaches a plateau

    Clifford R Jack et al. “Brain β-amyloid load approaches a plateau”. In: Neurology 80.10 (2013), pp. 890–896

    2013

  17. [25]

    A computational neurodegenerative disease progression score: method and results with the Alzheimer’s disease neuroimaging initiative cohort

    Bruno M Jedynak et al. “A computational neurodegenerative disease progression score: method and results with the Alzheimer’s disease neuroimaging initiative cohort”. In: Neuroimage 63.3 (2012), pp. 1478–1486

    2012

  18. [26]

    Simple video generation using neural ODEs

    David Kanaa et al. “Simple video generation using neural ODEs”. In: arXiv preprint arXiv:2109.03292 (2021)

    2021 arXiv

  19. [27]

    Diffusion models in medical imaging: A comprehensive survey

    Amirhossein Kazerouni et al. “Diffusion models in medical imaging: A comprehensive survey”. In: Medical Image Analysis 88 (2023), p. 102846

    2023

  20. [28]

    Medical Diffusion–Denoising Diffusion Probabilistic Models for 3D Medical Image Generation

    Firas Khader et al. “Medical Diffusion–Denoising Diffusion Probabilistic Models for 3D Medical Image Generation”. In: arXiv preprint arXiv:2211.03364 (2022)

    2022 arXiv

  21. [29]

    Transformers for CT reconstruction from monoplanar and biplanar radiographs

    Firas Khader et al. “Transformers for CT reconstruction from monoplanar and biplanar radiographs”. In: International Workshop on Simulation and Synthesis in Medical Imaging. Springer. 2023, pp. 1–10

    2023

  22. [30]

    Longitudinal brain MR image modeling using personal- ized memory for Alzheimer’s disease

    Seong Tae Kim, Umut Küçükaslan, and Nassir Navab. “Longitudinal brain MR image modeling using personal- ized memory for Alzheimer’s disease”. In: IEEE Access 9 (2021), pp. 143212–143221

    2021

  23. [31]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. “Adam: A method for stochastic optimization”. In: arXiv preprint arXiv:1412.6980 (2014)

    2014 arXiv

  24. [32]

    Auto-encoding variational Bayes

    Diederik P Kingma and Max Welling. “Auto-encoding variational Bayes”. In:arXiv preprint arXiv:1312.6114 (2013)

    2013 arXiv

  25. [33]

    CCVS: Context-aware controllable video synthesis

    Guillaume Le Moing, Jean Ponce, and Cordelia Schmid. “CCVS: Context-aware controllable video synthesis”. In: Advances in Neural Information Processing Systems. V ol. 34. 2021, pp. 14042–14055

    2021

  26. [34]

    Hi-BEHRT: Hierarchical Transformer-based model for accurate prediction of clinical events using multimodal longitudinal electronic health records

    Yikuan Li et al. “Hi-BEHRT: Hierarchical Transformer-based model for accurate prediction of clinical events using multimodal longitudinal electronic health records”. In: IEEE Journal of Biomedical and Health Informatics 27.2 (2022), pp. 1106–1117

    2022

  27. [35]

    3D brain and heart volume generative models: A survey

    Yanbin Liu et al. “3D brain and heart volume generative models: A survey”. In:ACM Computing Surveys 56.6 (2024), pp. 1–37. 7 A PREPRINT - F EBRUARY 11, 2025

    2024

  28. [36]

    Disentangling normal aging from Alzheimer’s disease in structural magnetic resonance images

    Marco Lorenzi et al. “Disentangling normal aging from Alzheimer’s disease in structural magnetic resonance images”. In: Neurobiology of Aging 36 (2015), S42–S52

    2015

  29. [37]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. “Decoupled weight decay regularization”. In:arXiv preprint arXiv:1711.05101 (2017)

    2017 arXiv

  30. [38]

    MC-RV AE: Multi-channel recurrent variational autoencoder for multimodal Alzheimer’s disease progression modelling

    Gerard Martí-Juan et al. “MC-RV AE: Multi-channel recurrent variational autoencoder for multimodal Alzheimer’s disease progression modelling”. In: NeuroImage 268 (2023), p. 119892

    2023

  31. [39]

    Learning imaging biomarker trajectories from noisy Alzheimer’s disease data using a Bayesian multilevel model

    Neil P Oxtoby et al. “Learning imaging biomarker trajectories from noisy Alzheimer’s disease data using a Bayesian multilevel model”. In: Bayesian and grAphical Models for Biomedical Imaging: First International Workshop, BAMBI 2014, Cambridge, MA, USA, September 18, 2014, Rev...

    2014

  32. [40]

    Vid-ODE: Continuous-time video generation with neural ordinary differential equation

    Sunghyun Park et al. “Vid-ODE: Continuous-time video generation with neural ordinary differential equation”. In: Proceedings of the AAAI Conference on Artificial Intelligence. V ol. 35. 3. 2021, pp. 2412–2422

    2021

  33. [41]

    Enhancing spatiotemporal disease progression models via latent diffusion and prior knowledge

    Lemuel Puglisi, Daniel C Alexander, and Daniele Ravı‘. “Enhancing spatiotemporal disease progression models via latent diffusion and prior knowledge”. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer. 2024, pp. 173–183

    2024

  34. [42]

    Degenerative adversarial neuroimage nets for brain scan simulations: Application in ageing and dementia

    Daniele Ravi et al. “Degenerative adversarial neuroimage nets for brain scan simulations: Application in ageing and dementia”. In: Medical Image Analysis 75 (2022), p. 102257

    2022

  35. [43]

    Generating diverse high-fidelity images with VQ-V AE-2

    Ali Razavi, Aaron Van den Oord, and Oriol Vinyals. “Generating diverse high-fidelity images with VQ-V AE-2”. In: Advances in Neural Information Processing Systems. V ol. 32. 2019

    2019

  36. [44]

    U-Net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. “U-Net: Convolutional networks for biomedical image segmentation”. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer. 2015, pp. 234–241

    2015

  37. [45]

    Latent ordinary differential equations for irregularly- sampled time series

    Yulia Rubanova, Ricky TQ Chen, and David K Duvenaud. “Latent ordinary differential equations for irregularly- sampled time series”. In: Advances in neural information processing systems. V ol. 32. 2019

    2019

  38. [46]

    Progression models for imaging data with Longitudinal Variational Auto Encoders

    Benoı‘t Sauty and Stanley Durrleman. “Progression models for imaging data with Longitudinal Variational Auto Encoders”. In: International Conference on Medical image computing and computer-assisted intervention. Springer. 2022, pp. 3–13

    2022

  39. [47]

    InterfaceGAN: Interpreting the disentangled face representation learned by GANs

    Yujun Shen et al. “InterfaceGAN: Interpreting the disentangled face representation learned by GANs”. In: IEEE Transactions on Pattern Analysis and Machine Intelligence44.4 (2020), pp. 2004–2018

    2020

  40. [48]

    Very deep convolutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. “Very deep convolutional networks for large-scale image recognition”. In: arXiv preprint arXiv:1409.1556 (2014)

    2014 arXiv

  41. [49]

    Medical image generation using generative adversarial networks: A review

    Nripendra Kumar Singh and Khalid Raza. “Medical image generation using generative adversarial networks: A review”. In: Health informatics: A computational perspective in healthcare (2021), pp. 77–96

    2021

  42. [50]

    A model of brain morphological changes related to aging and Alzheimer’s disease from cross-sectional assessments

    Raphaël Sivera et al. “A model of brain morphological changes related to aging and Alzheimer’s disease from cross-sectional assessments”. In: NeuroImage 198 (2019), pp. 255–270

    2019

  43. [51]

    Hierarchical amortized GAN for 3D high resolution medical image synthesis

    Li Sun et al. “Hierarchical amortized GAN for 3D high resolution medical image synthesis”. In: IEEE Journal of Biomedical and Health Informatics 26.8 (2022), pp. 3966–3975

    2022

  44. [52]

    Neuromorphologically-preserving volumetric data encoding using VQ-V AE

    Petru-Daniel Tudosiu et al. “Neuromorphologically-preserving volumetric data encoding using VQ-V AE”. In: arXiv preprint arXiv:2002.05692 (2020)

    2020 arXiv

  45. [53]

    Multi-scale GANs for memory-efficient generation of high resolution medical images

    Hristina Uzunova et al. “Multi-scale GANs for memory-efficient generation of high resolution medical images”. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer. 2019, pp. 112–120

    2019

  46. [54]

    Neural discrete representation learning

    Aaron Van Den Oord, Oriol Vinyals, et al. “Neural discrete representation learning”. In: Advances in neural information processing systems. V ol. 30. 2017

    2017

  47. [55]

    Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: a prospective cohort study

    Victor L Villemagne et al. “Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: a prospective cohort study”. In: The Lancet Neurology 12.4 (2013), pp. 357–367

    2013

  48. [56]

    MCVD-masked conditional video diffusion for prediction, generation, and interpolation

    Vikram V oleti, Alexia Jolicoeur-Martineau, and Chris Pal. “MCVD-masked conditional video diffusion for prediction, generation, and interpolation”. In: Advances in Neural Information Processing Systems . V ol. 35. Curran Associates, Inc., 2022, pp. 23371–23385

    2022

  49. [57]

    Modelling Latent Dynamics of StyleGAN using Neural ODEs

    Weihao Xia, Yujiu Yang, and Jing-Hao Xue. “Modelling Latent Dynamics of StyleGAN using Neural ODEs”. In: arXiv preprint arXiv:2208.11197 (2022)

    2022 arXiv

  50. [58]

    Videogpt: Video generation using VQ-V AE and transformers

    Wilson Yan et al. “Videogpt: Video generation using VQ-V AE and transformers”. In: arXiv preprint arXiv:2104.10157 (2021)

    2021 arXiv

  51. [59]

    SADM: Sequence-aware diffusion model for longitudinal medical image generation

    Jee Seok Yoon et al. “SADM: Sequence-aware diffusion model for longitudinal medical image generation”. In: International Conference on Information Processing in Medical Imaging. Springer. 2023, pp. 388–400. 8 A PREPRINT - F EBRUARY 11, 2025

    2023

  52. [60]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang et al. “The unreasonable effectiveness of deep features as a perceptual metric”. In: Proceedings of the IEEE conference on computer vision and pattern recognition. 2018, pp. 586–595. A Pre-processing The registration process aligns corresponding structures across...

    2018

  53. [61]

    VQ-GAN [10], a variant of VQ-V AE, incorporated a discriminator and perceptual loss, combining the strengths of both VQ-V AE and GAN to generate high-resolution images

    utilized hierarchical multi-scale latent maps for large-scale image generation. VQ-GAN [10], a variant of VQ-V AE, incorporated a discriminator and perceptual loss, combining the strengths of both VQ-V AE and GAN to generate high-resolution images. Ge et al. [16] extended VQ-G...

    2025

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.