Pith. sign in

REVIEW 5 major objections 6 minor 25 references

D2Diff : A Dual Domain Diffusion Model for Accurate Multi-Contrast MRI Synthesis

T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Conditioning diffusion on frequency and spatial features improves multi-contrast MRI synthesis.

desk verdict A plausible new dual-domain diffusion architecture for multi-contrast MRI synthesis, with real but modest gains over SynDiff; the main weakness is that the headline superiority claim is not backed by reported statistics. read the letter →

arxiv 2506.15750 v1 pith:PISX7K4Q submitted 2025-06-18 eess.IV

Sanuwani Dayarathna , Himashi Peiris , Kh Tohidul Islam , Tien-Tsin Wong , Zhaolin Chen This is my paper · ORCID
classification eess.IV
keywords multi-contrastMRIsynthesisdiffusionmodelsdual-domainlearningfrequencydomainfeaturesdiscretecosinetransformuncertainty-awaremasklossbraintumourimagingmedicalimage
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

Multi-contrast MRI synthesis asks a model to create one contrast, such as T1CE or FLAIR, from the other available contrasts. This paper claims that the missing contrast is better estimated when denoising is conditioned on both spatial features and frequency features at once, rather than on spatial features alone. It proposes D2Diff, a diffusion model with two jointly trained denoising networks—one fed multi-scale DCT frequency features, the other fed spatial residual features—plus a shared critic that aligns them and an uncertainty mask that focuses learning on difficult regions. On brain tumour and healthy datasets, D2Diff reports higher PSNR and SSIM and lower MAE than the compared methods, and tumour segmentation on synthetic images stays close to segmentation on real images. The point of the claim is that dual-domain conditioning yields clinically usable synthetic contrasts, reducing scan time and contrast-agent exposure.

What carries the argument

The load-bearing mechanism is dual-domain conditioning inside a denoising diffusion process. The frequency branch computes DCT features with kernel sizes 3, 5, and 7 for each source contrast, weights them with a softmax attention module, and fuses them into a conditioning vector; the spatial branch extracts per-contrast residual features. Two encoder-decoder denoisers then reverse the diffusion process, one guided by each branch, while a shared critic provides adversarial consistency and uncertainty maps. During sampling, the two networks' posterior means are averaged, so both domains contribute to every denoising step instead of being fused only at the input level.

What would settle it

Run the same training pipeline on the T1CE task with the frequency branch replaced by a second spatial branch of matched parameter count. If synthesis metrics do not drop meaningfully, the claim that frequency-domain features carry complementary information is not supported.

Watch

Extended reading notes

Core claim

The central discovery is that a diffusion model for multi-contrast MRI can be made more accurate by feeding two separate denoising networks complementary views of the conditioning contrasts: frequency-domain features extracted by discrete cosine transforms at three kernel scales, and spatial-domain residual features. The two networks are trained as a pair through a shared time-dependent critic, and at sampling their predicted denoised images are combined by averaging the posterior means. An uncertainty-aware mask loss derived from critic attention maps reweights the synthesis loss toward high-uncertainty regions. In the reported comparisons, the full D2Diff system outperforms the previous state-of-the-art baselines on all tested contrasts, and its synthetic images support downstream tumour segmentation with Dice scores comparable to those obtained from real contrasts.

Load-bearing premise

The load-bearing premise is that the hand-chosen DCT frequency features are genuinely complementary to spatial features; if replacing the frequency branch with another spatial branch of the same size changes nothing, the dual-domain design has no real advantage.

Editorial extensions

If this is right

  • If D2Diff works as reported, multi-contrast MRI synthesis improves across standard fidelity metrics, with the largest visible gains on challenging tumour-enhancement contrasts such as T1CE.
  • Downstream tumour segmentation using synthetic contrasts tracks segmentation on real contrasts, suggesting the synthetic images carry diagnostically relevant information rather than looking plausible only.
  • The mutual-learning design means each domain branch alone is weaker than the pair; ablation results show combined dual-domain training outperforms either individual branch.
  • The uncertainty mask loss and adaptive frequency aggregation each add measurable benefit, so the full architecture, not just the diffusion backbone, drives the result.

Reading between the lines

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

  • The fixed DCT kernel sizes of 3, 5, and 7 are an arbitrary choice; learning the kernels or using a learned frequency transform might yield further gains, especially for 3D volume synthesis.
  • Because frequency features encode global intensity shifts, the frequency branch may also make the model more robust to scanner-to-scanner intensity variation; the paper's multi-site data hints at this but does not test it explicitly.
  • The same dual-domain conditioning could plausibly transfer to other synthesis settings, such as CT from MRI or missing-modality PET, where global intensity differences between modalities are also prominent.
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

5 major / 6 minor

Summary. The paper proposes D2Diff, a dual-domain diffusion framework for multi-contrast MRI synthesis. The method trains two U-Net denoising networks jointly via a shared critic: one network is conditioned on spatial-domain features extracted from source contrasts, and the other on frequency-domain features obtained by applying multi-scale discrete cosine transforms. An uncertainty-aware mask loss derived from critic attention maps is used to focus learning on high-uncertainty regions, and inference averages the predictions of the two networks. The method is evaluated on BraTS2019 and a healthy brain dataset for synthesis of T1, T1CE, T2, and FLAIR contrasts, compared with GAN- and diffusion-based baselines, and further assessed through a downstream tumour segmentation task and an ablation study. The authors report consistent improvements over baselines on PSNR, SSIM, and MAE, and claim statistical significance at p<0.05.

Significance. If validated, the paper makes a useful contribution to multi-contrast MRI synthesis by explicitly combining spatial and frequency-domain conditioning inside a diffusion framework, with a shared critic for mutual learning and an uncertainty mask for region-focused training. The inclusion of a downstream segmentation evaluation and an ablation study is commendable, as is the stated intent to release code and hyperparameters. However, the central empirical claim of superiority over state-of-the-art baselines currently rests on a statistical assertion that is not substantiated, and several implementation details needed for reproducibility are missing. The work is therefore significant in scope and potential impact, but the evidence as presented is not yet sufficient to support the headline claims.

major comments (5)
  1. [Section 4, Table 1] The note 'statistical significance p<0.05 based on paired mean t-test' is not backed by any reported p-values, confidence intervals, or test details. Several D2Diff-vs-SynDiff margins are smaller than the reported standard deviations (e.g., BraTS T1CE PSNR 28.58±2.69 vs 28.16±2.36; BraTS FLAIR PSNR 27.57±2.18 vs 27.13±2.11), and with only 30 test subjects the significance of such small differences depends on the correlation of paired errors, which is not reported. Please provide the actual p-values with the test details for every comparison in Table 1, or temper the superiority claims to what the data support.
  2. [Section 2, Eq. (6) and Fig. 1B] The description 'DCT with kernel sizes 3, 5, and 7' is ambiguous. A discrete cosine transform is a global or blockwise transform; the term 'kernel size' suggests block-wise DCT with patch sizes 3×3, 5×5, and 7×7, but the paper does not specify patch overlap, whether coefficients are truncated or fully retained, or how the resulting 'frequency features' are structured. This ambiguity is load-bearing because the claimed complementarity of frequency-domain features is a central design choice, and the current text does not allow the operation to be reproduced or compared with other frequency-domain approaches. Please define the exact DCT implementation and, ideally, provide an ablation or reference justifying the chosen kernel sizes.
  3. [Section 3 and Section 4] The experimental setup omits essential training and architectural details: the total number of diffusion timesteps T, the noise schedule beta_t, the loss weights lambda_s, lambda_m, lambda_c, learning rate, batch size, number of training iterations, optimizer, and the specific U-Net configurations for both denoising networks and the critic. The abstract states that 'Code and model hyperparameters are available,' but the paper itself does not report them. Without these details, the fair comparison with baselines cannot be assessed and the experiments are not reproducible from the manuscript alone.
  4. [Section 4, Table 2] The downstream segmentation results are presented as Dice scores without standard deviations, confidence intervals, or significance tests. The differences are small (D2Diff 81.05 vs Complete 80.83; D2Diff 81.05 vs SynDiff 80.91), and the claim that 'superior downstream tumour segmentation highlights the diagnostic value of the synthetic images' is not supported without an estimate of variability. Please report error bars from multiple segmentation training runs or a statistical comparison, and adjust the interpretation accordingly.
  5. [Section 4, Table 3] The ablation study reports very small margins between D2Diff and the 'w/o freq. feat. adaptation' and 'Spatial guidance (H2)' variants (e.g., PSNR 28.58 vs 28.54 and 28.39; SSIM 91.84 vs 91.34 and 91.82; MAE 1.97 vs 2.07 and 2.02). Without significance testing or multiple runs, the claim that 'every component contributes to enhancing overall synthesis quality' is not established. Please provide statistical evidence or effect sizes for the ablation comparisons.
minor comments (6)
  1. [Section 2, Eq. (7)] The notation in Eq. (7) is difficult to parse: the expression 'f1 = [ Wi · sum ... ]' mixes a summation over k and a subscript i in a way that is not mathematically clean. Please rewrite this equation with clear indexing over the n source contrasts.
  2. [Section 2, Eq. (10)] The equations for the critic loss and adversarial loss contain unclear notation such as 'xpj t−1' and the use of eta; please clarify the dependence on j and define all variables before first use.
  3. [Section 3, Datasets] The healthy dataset description says 'we extracted 100 middle slices from 85 healthy brain MRI scans' and then 'We allocated 50, 20 and 15 subjects for training, validation, and testing.' This is inconsistent: slices are counted but subjects are allocated, and 50+20+15=85. Please clarify whether the splits are per subject or per slice and reconcile the numbers.
  4. [Section 4, Table 1] The table reports MAE values for some baselines (e.g., MM-GAN on T2 and T1) that are several times larger than the PSNR values would suggest; please check whether these values are percentages or normalized and state the metric definitions explicitly in a footnote.
  5. [References] Reference [17] is formatted incorrectly as 'Al., M.'; the correct citation is to Menze et al., and the reference should be updated. Also, reference [25] lists 'Others Pytorch' and should be cleaned up.
  6. [Section 4, Fig. 4] The qualitative segmentation figure would benefit from clearly visible tumor boundaries and a longer caption explaining what is shown in each column, especially since the claimed segmentation improvement is central to the diagnostic-value argument.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported gains rest on held-out test evaluation and external baselines, not on construction or load-bearing self-citation.

full rationale

The paper's central claim is that D2Diff outperforms existing methods on multi-contrast MRI synthesis. This is supported by quantitative metrics (PSNR, SSIM, MAE) computed on held-out test subjects in Table 1, by downstream segmentation Dice scores in Table 2, and by an ablation study in Table 3. None of these results is forced by the definitions or equations in Section 2: the loss functions in Eqs. (10)-(15) are standard supervised and adversarial objectives, and the inference rule in Eqs. (17)-(19) averages the two denoising networks' predictions rather than embedding the evaluation metric. The DCT-based frequency features in Eq. (6) are an architectural design choice, not a fitted parameter renamed as a prediction, and the uncertainty mask loss in Eqs. (12)-(13) is a training heuristic rather than a quantity that trivially equals the reported outcome. The paper does cite prior work by the same group (refs. [5], [19], [23]), but these citations provide background, a dataset source, and related review material; the superiority claim does not depend on any uniqueness theorem or unverified self-citation. The manuscript's statistical reporting is weak (the claimed p<0.05 is not accompanied by p-values, and several margins are within one standard deviation of the strongest baseline), but that is a correctness/evidence concern, not circularity. No step in the derivation reduces by construction to its own input, so the circularity score is 0.

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

This is an empirical deep learning paper. The central claim depends on standard diffusion and GAN machinery plus hand-chosen architectural choices like the DCT kernel sizes and loss weights. No new physical entities or formal axioms are introduced; the main burden is that the DCT-based frequency features are assumed complementary, and key hyperparameters are undisclosed.

free parameters (4)
  • Loss weights lambda_s, lambda_m, lambda_c
    Weights in Eq. 15 controlling supervised, mask, and critic loss terms; chosen by hand or validation, but values are not reported.
  • Number of diffusion timesteps T
    Reverse sampling runs T steps (Eq. 16-19); T is not stated.
  • Noise schedule beta_t
    Variance schedule in forward diffusion (Eq. 2); parameters not specified.
  • DCT kernel sizes = 3, 5, 7
    Chosen by hand in Eq. 6 without an experiment isolating each size.
assumptions (4)
  • standard math Diffusion forward and reverse process parameterization from Ho et al. 2020.
    Used in Eqs. 2-4 and Eq. 16 as the backbone of the synthesis method.
  • standard math Denoising Diffusion GAN critic and adversarial training from Xiao et al. 2021.
    The shared critic and adversarial losses (Eqs. 10-11) are adopted from Ref. [13].
  • domain assumption Source contrasts y_i are co-registered and contain sufficient information to synthesize the target contrast.
    Problem formulation in Section 2 assumes paired, aligned multi-contrast images; registration errors would degrade synthesis.
  • ad hoc to paper Small-block DCT coefficients provide meaningful frequency-domain features for MRI synthesis.
    The paper's core modeling premise, stated in Section 2 with no independent justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of D2Diff : A Dual Domain Diffusion Model for Accurate Multi-Contrast MRI Synthesis." pith.science (2026). https://pith.science/paper/PISX7K4Q

@misc{pith2026250615750,
  author       = {Pith},
  title        = {Pith review of: D2Diff : A Dual Domain Diffusion Model for Accurate Multi-Contrast MRI Synthesis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PISX7K4Q}},
  note         = {Machine review of arXiv:2506.15750}
}
read the original abstract

Multi contrast MRI synthesis is inherently challenging due to the complex and nonlinear relationships among different contrasts. Each MRI contrast highlights unique tissue properties, but their complementary information is difficult to exploit due to variations in intensity distributions and contrast specific textures. Existing methods for multi contrast MRI synthesis primarily utilize spatial domain features, which capture localized anatomical structures but struggle to model global intensity variations and distributed patterns. Conversely, frequency domain features provide structured inter contrast correlations but lack spatial precision, limiting their ability to retain finer details. To address this, we propose a dual domain learning framework that integrates spatial and frequency domain information across multiple MRI contrasts for enhanced synthesis. Our method employs two mutually trained denoising networks, one conditioned on spatial domain and the other on frequency domain contrast features through a shared critic network. Additionally, an uncertainty driven mask loss directs the models focus toward more critical regions, further improving synthesis accuracy. Extensive experiments show that our method outperforms SOTA baselines, and the downstream segmentation performance highlights the diagnostic value of the synthetic results.

Figures

Figures reproduced from arXiv: 2506.15750 by the authors.

Figure 1
Figure 1. Network architecture. A: Overall Architecture with frequency (ϕ1) and spatial guidance (ϕ2), B: Multi-scale adaptive frequency aggregation in ϕ1. Dual Domain Diffusion Model [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Visualization of synthetic MRI results on healthy dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Visualization of synthetic MRI results on BraTS dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Segmentation results. Task Contrasts Dice Score% ↑ T1 T2 FLAIR T1CE pGAN Hi-Net SynDiff D2Diff a ✓ x ✓ ✓ 80.5 80.02 80.91 81.05 b ✓ x ✓ x 80.64 79.70 80.81 81.19 c x x ✓ x 79.96 78.93 80.66 81.02 d x x x x 76.50 73.49 79.31 79.34 complete ✓ ✓ ✓ ✓ 80.83 [PITH_FULL_IMAG…

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 12 canonical work pages

  1. [1]

    Wei, K., Kong, W., Liu, L., Wang, J., Li, B., Zhao, B., Li, Z., Zhu, J. & Yu, G. CT synthesis from MR images using frequency attention conditional generative adversarial network.Computers In Biology And Medicine.170pp. 107983 (2024,3), https://doi.org/10.1016/j.compbiomed.2024.107983

    work page arXiv 2024

  2. [2]

    & Rao, K

    Ahmed, N., Natarajan, T. & Rao, K. Discrete Cosine Transform.IEEE Transactions On Computers.C-23, 90-93 (1974)

    work page 1974

  3. [3]

    & Avestimehr, S

    Ziashahabi, A., Buyukates, B., Sheshmani, A., You, Y. & Avestimehr, S. Frequency Domain Diffusion Model with Scale-Dependent Noise Schedule.2024 IEEE Inter- national Symposium On Information Theory (ISIT). pp. 19-24 (2024)

    work page 2024

  4. [4]

    & Abbeel, P

    Ho, J., Jain, A. & Abbeel, P. Denoising Diffusion Probabilistic Models.Advances In Neural Information Processing Systems.33pp. 6840-6851 (2020)

    work page 2020

  5. [5]

    & Chen, Z

    Dayarathna, S., Islam, K., Uribe, S., Yang, G., Hayat, M. & Chen, Z. Deep learning based synthesis of MRI, CT and PET: Review and analysis.Medical Image Analysis. 92pp. 103046 (2024,2), https://doi.org/10.1016/j.media.2023.103046

    arXiv 2024

  6. [6]

    & Pietsch, H

    Lohrke, J., Frenzel, T., Endrikat, J., Alves, F., Grist, T., Law, M., Lee, J., Leiner, T., Li, K., Nikolaou, K., Prince, M., Schild, H., Weinreb, J., Yoshikawa, K. & Pietsch, H. 25 Years of Contrast-Enhanced MRI: Developments, Current Challenges and Future Perspectives.Advances In Therapy.33, 1-28 (2016,1), https://doi.org/10. 1007/s12325-015-0275-4

    work page 2016

  7. [7]

    & Gabr, R

    Narayana, P., Coronado, I., Sujit, S., Sun, X., Wolinsky, J. & Gabr, R. Are multi- contrast magnetic resonance images necessary for segmenting multiple sclerosis brains? A large cohort study based on deep learning.Magnetic Resonance Imag- ing.65pp. 8-14 (2020,1), https://doi.org/10.1016/j.mri.2019.10.003

    work page doi:10.1016/j.mri.2019.10.003 2020

  8. [8]

    & Wang, Y

    Zhan, B., Li, D., Wu, X., Zhou, J. & Wang, Y. Multi-Modal MRI Image Synthesis via GAN With Multi-Scale Gate Mergence.IEEE Journal Of Biomedical And Health Informatics.26, 17-26 (2022,1), https://doi.org/10.1109/JBHI.2021.3088866

    arXiv 2022

Show all 25 references
  1. [9]

    & Shao, L

    Zhou, T., Fu, H., Chen, G., Shen, J. & Shao, L. Hi-Net: Hybrid-Fusion Network for Multi-Modal MR Image Synthesis.IEEE Transactions On Medical Imaging.39pp. 2772-2781 (2020), https://api.semanticscholar.org/CorpusID:211082884

    2020

  2. [10]

    & Undefinedukur, T

    Yurt, M., Dar, S., Erdem, A., Erdem, E., Oguz, K. & Undefinedukur, T. mustGAN: multi-stream Generative Adversarial Networks for MR Image Synthesis.Medical Im- age Analysis.70pp. 101944 (2021,5), https://doi.org/10.1016/j.media.2020.101944

    2021

  3. [11]

    & Zhou, L

    Zhan, B., Li, D., Wang, Y., Ma, Z., Wu, X., Zhou, J. & Zhou, L. LR-cGAN: Latent representation based conditional generative adversarial network for multi-modality MRI synthesis.Biomedical Signal Processing And Control.66pp. 102457 (2021,4), https://doi.org/10.1016/j.bspc.2021.102457

    2021

  4. [12]

    & Hamarneh, G

    Sharma, A. & Hamarneh, G. Missing MRI Pulse Sequence Synthesis Using Multi- Modal Generative Adversarial Network.IEEE Transactions On Medical Imaging. 39, 1170-1183 (2020,4), https://doi.org/10.1109/TMI.2019.2945521

    2020

  5. [13]

    & Vahdat, A

    Xiao, Z., Kreis, K. & Vahdat, A. Tackling the Generative Learning Trilemma with Denoising Diffusion GANs.ArXiv.abs/2112.07804(2021), https://api.semanticscholar.org/CorpusID:245144350

    2021 arXiv

  6. [14]

    & Efros, A

    Isola, P., Zhu, J., Zhou, T. & Efros, A. Image-to-Image Translation with Con- ditional Adversarial Networks.2017 IEEE Conference On Computer Vision And Pattern Recognition (CVPR). (2017,7), https://doi.org/10.1109/CVPR.2017.632

    2017 doi

  7. [15]

    & Cukur, T

    Dar, S., Yurt, M., Karacan, L., Erdem, A., Erdem, E. & Cukur, T. Image Synthesis in Multi-Contrast MRI With Conditional Generative Adversarial Net- works.IEEE Transactions On Medical Imaging.38, 2375-2388 (2019,10), https: //doi.org/10.1109/TMI.2019.2901750 10 S. Dayarathna et al

    2019

  8. [16]

    & Çukur, T

    Ozbey, M., Dalmaz, O., Dar, S., Bedel, H., Ozturk, S., Gungor, A. & Çukur, T. Unsupervised Medical Image Translation With Adversarial Diffusion Models.IEEE Transactions On Medical Imaging.42, 3524-3539 (2023)

    2023

  9. [17]

    The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)

    Al., M. The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS). IEEE Transactions On Medical Imaging.34, 1993-2024 (2015,10), https://doi.org/ 10.1109/tmi.2014.2377694

    1993

  10. [18]

    & Lenga, M

    Dohmen, M., Klemens, M., Baltruschat, I., Truong, T. & Lenga, M. Similarity and quality metrics for MR image-to-image translation.Scientific Reports.15(2025,1), https://doi.org/10.1038/s41598-025-87358-0

    2025 doi

  11. [19]

    & Chen, Z

    Islam, K., Zhong, S., Zakavi, P., Chen, Z., Kavnoudias, H., Farquharson, S., Dur- bridge, G., Barth, M., McMahon, K., Parizel, P., Dwyer, A., Egan, G., Law, M. & Chen, Z. Improving portable low-field MRI image quality through image-to-image translation using paired low- and hi...

    2023 doi

  12. [20]

    & Huang, Y

    Lou, Y., Zhang, J., Xu, D., Cao, Y., Wang, H. & Huang, Y. No-Reference MRI Quality Assessment via Contrastive Representation: Spatial and Frequency Do- main Perspectives.2024 IEEE International Conference On Multimedia And Expo (ICME). pp. 1-6 (2024,7),https://doi.org/10.1109/...

    2024

  13. [21]

    & Zhang, Y

    Ding, H., Lu, J., Cai, J. & Zhang, Y. SLf-UNet: Improved UNet for Brain MRI Seg- mentation by Combining Spatial and Low-Frequency Domain Features. (Research Square Platform LLC,2023,5), https://doi.org/10.21203/rs.3.rs-2849524/v1

    2023 doi

  14. [22]

    Cardoso, M. et. al., A. MONAI: An open-source framework for deep learning in healthcare. (arXiv,2022), https://arxiv.org/abs/2211.02701

    2022 arXiv

  15. [23]

    & Egan, G

    Chen, Z., Pawar, K., Ekanayake, M., Pain, C., Zhong, S. & Egan, G. Deep Learning for Image Enhancement and Correction in Magnetic Resonance Imaging—State-of- the-Art and Challenges.Journal Of Digital Imaging.36, 204-230 (2022,11), https: //doi.org/10.1007/s10278-022-00721-9

    2022 doi

  16. [24]

    & Zhou, L

    Yu, B., Wang, Y., Wang, L., Shen, D. & Zhou, L. Medical Image Synthesis via Deep Learning.Deep Learning In Medical Image Analysis. pp. 23-44 (2020), https: //doi.org/10.1007/978-3-030-33128-3_2

    2020 doi

  17. [25]

    & Others Pytorch: An imperative style, high-performance deep learning library.Advances In Neural Information Processing Systems.32(2019)

    Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L. & Others Pytorch: An imperative style, high-performance deep learning library.Advances In Neural Information Processing Systems.32(2019)

    2019

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.