Pith. sign in

REVIEW 3 major objections 5 minor 23 references

A predicted intensity ceiling improves virtual contrast breast MRI on all eight metrics.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

A shared, source-predicted intensity coordinate for latent breast MRI virtual contrast enhancement improves eight internal-cohort quality metrics over fixed and separate coordinate baselines.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A clearly-posed intensity-calibration fix for latent breast MRI VCE, with one honest limitation: the headline gains all come from the cohort that guided development. the 3 major comments →

arxiv 2608.03612 v1 pith:GQZTEPNT submitted 2026-08-04 eess.IV cs.CV

Predictive Enhancement Calibration for Latent Breast MRI Virtual Contrast Enhancement

Qin Lei , Hao Wu This is my paper
classification eess.IV cs.CV
keywords virtual contrast enhancementbreast MRIlatent flow transformerintensity calibrationradiomicspredictive enhancement calibrationDCE-MRIimage synthesis
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

Virtual contrast enhancement (VCE) aims to synthesize a post-contrast breast MRI from a pre-contrast scan, avoiding the need for gadolinium. The paper argues that the main barrier to using modern pretrained latent image generators for this task is not the network but the intensity coordinate: MRI has no fixed intensity unit, while a frozen natural-image autoencoder expects bounded inputs. It proposes Predictive Enhancement Calibration (PEC), which places each pre-contrast/peak-contrast pair in a shared, case-adaptive interval during training and, at inference, predicts the unavailable upper endpoint from source-image statistics. On the fixed MAMA100 development cohort, source-only PEC improves all eight evaluation metrics over matched fixed-window and separate-coordinate baselines. If correct, this makes strong pretrained latent generators usable for MRI synthesis with only a small trainable adapter.

Core claim

The paper's central claim is that intensity mapping is a learnable interface: the upper endpoint of the encoding window changes radiomic distances before generation, and independent source/target scaling assigns different physical meanings to equal coordinate values. PEC encodes both images into one case-adaptive interval [ℓ(x), u(y)], with ℓ(x) the pre-contrast finite minimum and u(y) the 99.99th percentile of the target; an FT-Transformer predicts û(x) from 47 source statistics, so deployment needs neither target nor tumor mask. Oracle PEC reaches the lowest round-trip Fréchet Radiomic Distance (1.280); source-predicted PEC degrades to 9.186, exposing tail clipping. In conditional generati

What carries the argument

The key object is the PEC coordinate transform: C(z) = clip(((z − ℓ)/(u − ℓ))^(1/γ), 0, 1) with γ = 2.2, and its inverse D(a) = ℓ + (u − ℓ) a^γ. It is 'shared' because one interval [ℓ, u] is used for both source encoding and output decoding; 'case-adaptive' because ℓ comes from the pre-contrast slice and u from the target's 99.99th percentile; and 'predictive' because at inference u is replaced by an FT-Transformer estimate û(x) built from 47 source-only statistics with asymmetric and monotonicity losses. The gamma exponent allocates more 8-bit code levels to the densely occupied lower intensity range, while the 99.99th-percentile endpoint preserves the sparse enhancement tail. Around this c

Load-bearing premise

The central claim rests on the assumption that iterative design on the fixed MAMA100 cohort did not inflate PEC's advantage over the fixed-wide baseline in the same-cohort evaluation.

What would settle it

Run the near-matched conditional comparison (fixed-wide, shared PEC, separate coordinates) on a held-out cohort not used in development, and test whether PEC's MSE and LPIPS paired differences remain negative with confidence intervals excluding zero; additionally, if replacing the predicted upper endpoint with the oracle endpoint does not substantially close the FRD gap (from 9.186 toward 1.280), the source-predictive interface itself is the weak point.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Intensity calibration should be treated as part of the generative model: a fixed upper endpoint that clips the enhancement tail changes FRD before synthesis, so future VCE systems should report calibration choices.
  • A pretrained natural-image latent generator can be reused for breast MRI VCE with only a rank-128 LoRA adapter plus a small endpoint predictor, avoiding expensive medical-codec training.
  • Equivalent-pair coordinates make source and target values comparable at inference; separate source/target scaling is a measurable handicap (MSE 0.9096 vs 0.7493 for PEC).
  • The endpoint predictor reaches high correlation (0.975) but still leaves tail clipping; improving tail fidelity is the remaining bottleneck for radiomic-level synthesis.
  • Lesion-level gains (Dice, HD95, tumor-ROI AUROC) are point-estimate improvements with wide confidence intervals; the strongest paired evidence is on global image metrics MSE and LPIPS.

Where Pith is reading between the lines

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

  • The same shared-predictive-coordinate idea could transfer to other MRI-to-MRI tasks (e.g., T1-to-T2 synthesis) and other latent generators, since the conflict is between MRI's non-canonical scale and any bounded autoencoder.
  • A stronger endpoint predictor—perhaps predicting a full quantile curve or using a generative model of the tail—could close the gap between predicted-window FRD (9.186) and oracle PEC (1.280), likely improving the conditional FRD as well.
  • Because MAMA100 informed development, the honest test of the method is a preregistered or external-cohort comparison; the paper's own sensitivity control (oracle encoding, predicted decoding) offers a cheap way to separate window capacity from predictor error on new data.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper addresses a real interface problem for latent breast MRI virtual contrast enhancement: pretrained natural-image autoencoders require bounded inputs, while MRI intensities are non-canonical and patient-specific. The authors propose Predictive Enhancement Calibration (PEC), which uses a shared, case-adaptive intensity interval [ℓ(x), u(y)] for source and target during training, with the target's upper endpoint predicted from source statistics at inference. They integrate PEC with a frozen FLUX.2 latent flow transformer via LoRA reference conditioning and an ROI-weighted flow loss. On a fixed internal 100-patient development cohort (MAMA100), they compare PEC against fixed-wide and separate-coordinate baselines under near-matched training budgets. They report that PEC improves all eight point estimates, with paired bootstrap confidence intervals excluding zero for MSE and LPIPS, while SSIMt, Dice, and HD95 intervals include zero and FRD/AUROCs are cohort-level point estimates. The paper also uses target round trips to separate representation loss from synthesis error and includes a sensitivity control for endpoint-scale error.

Significance. If the result holds, PEC is an appealingly simple and portable calibration layer that could be applied to other latent generative models for medical imaging. The paper is methodical in isolating representation effects before generation, in matching training budgets, and in reporting which confidence intervals exclude zero. The release of code is a further strength. However, the central radiomics motivation is not yet statistically supported: the only metrics with paired CIs excluding zero are global reconstruction metrics, whereas the headline radiomic fidelity metric (FRD) is a cohort-level point estimate. The evaluation cohort is explicitly stated to have informed development, so the risk of selection inflation is real. The idea is sound and worth pursuing, but the evidence as presented is too fragile for acceptance without additional validation.

major comments (3)
  1. [Section 6 / Table 2] The development and evaluation cohort are the same: the authors state 'MAMA100 informed development under archived standardization' and all comparative results are reported on this fixed 100-patient cohort. Because iterative choices (top-eight slice sampling, Q99.99 endpoint, gamma=2.2, ROI-loss weight λ=4, FT-Transformer predictor, final-epoch selection) plausibly used MAMA100 metrics, the point-estimate improvements in Table 2 could reflect selection on the evaluation cohort. This is especially consequential for FRD, the metric most aligned with the paper's radiomic-fidelity motivation, but only MSE and LPIPS have paired CIs excluding zero; FRD and both AUROCs are cohort-level point estimates without uncertainty intervals. The internal evidence is coherent but insufficient. Please provide an external frozen-pipeline evaluation, or at minimum a prespecified development/evaluation split
  2. [Table 1 vs. Table 2 / Section 5.1] There is a tension between the representation-level and synthesis-level results. Table 1 shows source-predicted PEC round trips have FRD 9.186 and FRD_VAE 9.632, markedly worse than oracle PEC (1.280/2.793) and even fixed-wide (1.663/3.133). Yet Table 2 shows conditional PEC synthesis has FRD 4.429, better than fixed-wide 4.838. If the shared predicted coordinate is the mechanism, one would expect the predicted coordinate's poor representation fidelity to hamper synthesis, not improve it. The sensitivity control (oracle encoding, predicted decoding) helps separate endpoint-scale error from tail clipping, but the conditional result still needs explanation: why does a coordinate that alone degrades radiomic distance produce better synthetic FRD? Please analyze this interaction, e.g., report oracle-coordinate conditional generation as an upper bound, and examine whether the FRD gain is driv
  3. [Section 3.2 / Section 4.1] The endpoint predictor fφ uses 47 hand-selected source statistics and is selected on only 20 held-out Yunnan cases. The reported endpoint MAE 1.371 and Pearson correlation 0.975 come from this small selection set, so the predictive performance may be optimistic. With 47 free statistics and no ablation, it is unclear which features are necessary and whether the predictor is overfitting the small selection set. Please report cross-validated endpoint prediction performance on a larger source-only set (e.g., held-out MAMA training-pool patients), include error bars, and justify the choice of the 47 statistics or provide an ablation.
minor comments (5)
  1. [Eq. (4)] The source statistics s(x) are not formally defined; please specify the 47 features and the exact loss terms ('asymmetric and monotonicity terms') used for the predictor.
  2. [Section 4.1] The term 'MAMA training patients' is ambiguous: clarify whether this means the 1,406-patient training pool, and specify which normalization statistics (e.g., z-score mean/std) are computed on which data.
  3. [Fig. 2] The labels 'PEC gain I' and 'PEC gain II' are not explained in the caption or text; please define how these cases were selected and what the labels indicate.
  4. [Section 3.2 / Eq. (1)] The notation Q99.99(y) is computed over finite pixels in the original field of view, but it is worth clarifying whether background air values are included before any masking and how this interacts with padding.
  5. [References] Reference [21] could be cited with more detailed metadata (dataset name, version, access date) so readers can locate the Yunnan cohort used for predictor selection.

Circularity Check

0 steps flagged

No circularity: endpoint prediction and shared-coordinate comparison are out-of-sample and not forced by construction.

full rationale

The paper's core claims do not reduce to their inputs. The PEC upper endpoint u(y)=Q99.99(y) is a target-derived label used only during training; the deployed predictor f_phi is trained on MAMA training patients and 80 Yunnan cases, selected on 20 held-out Yunnan cases, and evaluated on MAMA100, so the Table 2 comparison is genuine out-of-sample prediction rather than fitting to the test cohort. The fixed-wide and separate-coordinate baselines have no fitted parameters, so PEC's point-estimate improvement (e.g., MSE 0.812 to 0.749, FRD 4.838 to 4.429) is not forced by construction; it is an empirical result. The motivation that tail clipping degrades FRD is measured with an external metric (FRD) and used to justify the Q99.99 choice, but that is design motivation, not a circular derivation. The self-reported limitation 'FRD is cohort-level. MAMA100 informed development under archived standardization' (Sec. 6) is a genuine external-validity risk: hyperparameter and design choices may have been influenced by the same 100-patient cohort used for evaluation, and FRD/AUROC are cohort-level point estimates without paired CIs. That concern belongs to correctness/validity, not circularity, because no test-set label is used to fit the predictor or the generator. There are no load-bearing self-citations; the cited prior work (MAMA-SYNTH, Yunnan data, FLUX, EasyControl) supplies data, backbone, and conditioning mechanism, not the claimed result.

Axiom & Free-Parameter Ledger

5 free parameters · 6 axioms · 0 invented entities

The ledger lists the hand-set constants and fitted components that the PEC result depends on. The gamma exponent, ROI loss weight, percentile choice, slice-sampling cap, and the learned endpoint predictor all enter the pipeline directly; changing them changes the reported numbers. Background axioms cover the frozen FLUX.2 basis, the clinical target definition, mask reliability, sufficiency of the 47 statistics, representativeness of the internal cohort, and parity of the three conditional runs. No invented physical entities are introduced.

free parameters (5)
  • gamma (γ) = 2.2
    In Eqs. (2)-(3), the gamma exponent allocates code levels to the low/mid range before 8-bit quantization; chosen by hand, and its value affects how many levels the enhancement tail gets.
  • Upper-tail percentile Q99.99 = 99.99%
    In Eq. (1), the robust upper endpoint is the 99.99th percentile of the target. The paper's own sweep (Q99.5 to Q99.99) shows FRD varies from 10.598 to 0.090, so this choice materially drives radiomic fidelity.
  • ROI loss weight λ = 4
    In Eq. (5), λ=4 weights the tumour-mask region in the flow loss; chosen by hand and affects lesion-region training emphasis.
  • Endpoint predictor fφ (FT-Transformer, 128-dim, 4 blocks) = trained weights; endpoint MAE 1.371, ρ=0.975
    The learned mapping from 47 source statistics to the target endpoint is fitted to labels from the MAMA training pool and 80 Yunnan cases; its quality directly controls the inference coordinate and thus PEC's round-trip FRD (9.186 vs oracle 1.280).
  • Slice sampling cap (top-eight by mask area) = 8 slices
    Sec. 4.2: training data keeps at most eight tumour-positive slices ranked by mask area; the sampling policy was changed from the earlier broader strategy, and it changes the training distribution.
axioms (6)
  • domain assumption The frozen FLUX.2 natural-image autoencoder's bounded input domain is treated as fixed; the generator's latent basis cannot be changed.
    Secs. 3.1-3.3: the whole PEC design assumes the pretrained codec and 9B transformer are retained and only LoRA parameters are trained.
  • domain assumption The clinical target is the slice at the largest tumour-mask cross-section of the peak-enhancement post-contrast phase.
    Sec. 4.1: this target definition follows MAMA-SYNTH [13]; if the target definition is wrong, the 'enhancement' being predicted is mis-specified.
  • domain assumption Tumour masks used in the ROI-weighted loss are reliable at training.
    Sec. 3.3 Eq. (5): the mask controls the loss; mask errors propagate into the conditional generator.
  • ad hoc to paper 47 hand-selected source statistics are sufficient to predict the target upper endpoint.
    Sec. 3.2 Eq. (4): the feature set (quantiles, tail means, top-k, threshold fractions, moments, shape) is chosen for this paper; no optimality or sufficiency argument is given.
  • domain assumption MAMA100 is representative of the eventual clinical population, and development on it does not invalidate evaluation on it.
    Secs. 4.1 and 6: all results are on the fixed internal 100-patient cohort that informed development; external cohorts are explicitly listed as future tests.
  • domain assumption The flow-matching objective with 20 epochs and rank-128 LoRA converges to a good conditional distribution for all coordinate schemes equally.
    Sec. 4.3: comparisons assume 'comparable training budgets and backbone settings' make the three runs matched; final-epoch selection is used, but convergence parity is asserted, not measured.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Predictive Enhancement Calibration for Latent Breast MRI Virtual Contrast Enhancement." pith.science (2026). https://pith.science/paper/GQZTEPNT

@misc{pith2026260803612,
  author       = {Pith},
  title        = {Pith review of: Predictive Enhancement Calibration for Latent Breast MRI Virtual Contrast Enhancement},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GQZTEPNT}},
  note         = {Machine review of arXiv:2608.03612}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Virtual contrast enhancement (VCE) synthesizes enhanced breast MR images from pre-contrast acquisitions. Modern latent generators offer strong image priors, but their bounded natural-image autoencoders conflict with the non-canonical intensity scale of MRI. We show that the upper bound can alter radiomic fidelity before generation, while scaling source and target independently creates a coordinate inconsistency. We propose Predictive Enhancement Calibration (PEC), which represents each pair in a shared, case-adaptive coordinate during training and predicts its unavailable upper endpoint from the pre-contrast image at inference. We integrate PEC with a pretrained FLUX latent flow transformer via parameter-efficient reference conditioning. Target round trips first isolate representation loss before generation; near-matched conditional models then compare PEC with fixed-wide and separate coordinates under comparable training budgets and backbone settings. On the fixed internal MAMA100 development cohort, PEC improves all eight point estimates in this source-only VCE setting, with paired evidence strongest for MSE and LPIPS.\noindent\textbf{Code:} https://github.com/tanlei0/pec-breast-mri-vce

Figures

Figures reproduced from arXiv: 2608.03612 by Hao Wu, Qin Lei.

Figure 1
Figure 1. Figure 1: PEC–FLUX.2 architecture. (a) Target endpoints supervise a source-statistics transformer that predicts uˆ(x). (b) Generator training maps the pre-/peak-contrast pair to the shared coordinate [ℓ(x), u(y)], encodes the resulting 8-bit RGB inputs with the frozen VAE, and learns conditional LoRA with an ROI-weighted flow loss. (c) Source-only inference reuses [ℓ(x), uˆ(x)] for forward PEC and output inversion; … view at source ↗
Figure 2
Figure 2. Figure 2: Qualitative MAMA100 comparison of a representative and two PEC-gain cases. Columns show the input, acquired target, separate-coordinate and PEC predictions, and matched crops; post-contrast panels share one coordinate. High endpoint correlation can still hide tail clipping and code-allocation errors in predicted-window round trips; the sensitivity control separates window capacity from inverse-scale error.… view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

23 extracted references · 13 canonical work pages

  1. [1]

    https://github.com/black-forest-labs/ flux2(2026)

    Black Forest Labs: FLUX.2 [klein]. https://github.com/black-forest-labs/ flux2(2026)

    work page 2026

  2. [2]

    Radiology306(3), e213199 (2023).https: //doi.org/10.1148/radiol.213199

    Chung, M., Calabrese, E., Mongan, J., Ray, K.M., Hayward, J.H., Kelil, T., Sieberg, R., Hylton, N., Joe, B.N., Lee, A.Y.: Deep learning to simulate contrast-enhanced breast MRI of invasive breast cancer. Radiology306(3), e213199 (2023).https: //doi.org/10.1148/radiol.213199

    work page doi:10.1148/radiol.213199 2023

  3. [3]

    In: Proceedings of the 41st International Conference on Machine Learning

    Esser, P., Kulal, S., Blattmann, A., Entezari, R., Müller, J., Saini, H., Levi, Y., Lorenz, D., Sauer, A., Boesel, F., Podell, D., Dockhorn, T., English, Z., Rombach, R.: Scaling rectified flow transformers for high-resolution image synthesis. In: Proceedings of the 41st International Conference on Machine Learning. Proceedings of Machine Learning Researc...

    work page 2024

  4. [4]

    Sci- entific Data12(1), 453 (2025).https://doi.org/10.1038/s41597-025-04707-4

    Garrucho, L., Kushibar, K., Reidel, C.A., Joshi, S., Osuala, R., Tsirikoglou, A., Bobowicz, M., del Riego, J., Catanese, A., et al.: MAMA-MIA: A large-scale multi- center breast cancer DCE-MRI benchmark dataset with expert segmentations. Sci- entific Data12(1), 453 (2025).https://doi.org/10.1038/s41597-025-04707-4

    doi:10.1038/s41597-025-04707-4 2025

  5. [5]

    In: Advances in Neural Information Processing Systems

    Gorishniy, Y., Rubachev, I., Khrulkov, V., Babenko, A.: Revisiting deep learning models for tabular data. In: Advances in Neural Information Processing Systems. vol. 34, pp. 18932–18943 (2021),https://proceedings.neurips.cc/paper/2021/ hash/9d86d83f925f2149e9edb0ac3b49229c-Abstract.html

    work page 2021

  6. [6]

    In: International Con- ference on Learning Representations (2022),https://openreview.net/forum?id= nZeVKeeFYf9

    Hu, E.J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., Chen, W.: LoRA: Low-rank adaptation of large language models. In: International Con- ference on Learning Representations (2022),https://openreview.net/forum?id= nZeVKeeFYf9

    2022

  7. [7]

    In: Artificial Intelligence and Imaging for Diagnostic and Treatment Challenges in Breast Care

    Ibarra, S., del Riego, J., Catanese, A., Cuba, J., Cardona, J., Leon, N., Infante, J., Lekadir, K., Diaz, O., Osuala, R.: Comparing conditional diffusion models for synthesizing contrast-enhanced breast MRI from pre-contrast images. In: Artificial Intelligence and Imaging for Diagnostic and Treatment Challenges in Breast Care. pp. 226–236. Lecture Notes i...

    work page doi:10.1007/978-3-032-05559-0_23 2026

  8. [8]

    Medical Image Analysis110, 103943 (2026)

    Konz, N., Osuala, R., Verma, P., Chen, Y., Gu, H., Dong, H., Chen, Y., Marshall, A., Garrucho, L., et al.: Fréchet radiomic distance (FRD): A versatile metric for comparing medical imaging datasets. Medical Image Analysis110, 103943 (2026). https://doi.org/10.1016/j.media.2026.103943

    arXiv 2026

  9. [9]

    In: The Eleventh International Conference on Learning Representa- tions

    Lipman, Y., Chen, R.T.Q., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for gen- erative modeling. In: The Eleventh International Conference on Learning Representa- tions. OpenReview.net (2023),https://openreview.net/forum?id=PqvMRDCJT9t

    work page 2023

  10. [10]

    Radiology292(3), 520–536 (2019).https://doi.org/10.1148/radiol.2019182947

    Mann, R.M., Cho, N., Moy, L.: Breast MRI: State of the art. Radiology292(3), 520–536 (2019).https://doi.org/10.1148/radiol.2019182947

    doi:10.1148/radiol.2019182947 2019

  11. [11]

    Radiology307(3), e222211 (2023).https://doi.org/10.1148/ radiol.222211

    Müller-Franzes, G., Huck, L., Tayebi Arasteh, S., Khader, F., Han, T., Schulz, V., Dethlefsen, E., Kather, J.N., Nebelung, S., Nolte, T., Kuhl, C., Truhn, D.: Using machine learning to reduce the need for contrast agents in breast MRI through synthetic images. Radiology307(3), e222211 (2023).https://doi.org/10.1148/ radiol.222211

    work page 2023

  12. [12]

    IEEE Transactions on Medical Imaging19(2), 143–150 (2000)

    Nyúl, L.G., Udupa, J.K., Zhang, X.: New variants of a method of MRI scale standardization. IEEE Transactions on Medical Imaging19(2), 143–150 (2000). https://doi.org/10.1109/42.836373 10 Q. Lei and H. Wu

    doi:10.1109/42.836373 2000

  13. [13]

    Osuala, R., Joshi, S., van Dijk, J., Han, L., Cosaka, M.L., Mysler, D., Garrucho, L., Lekadir, K., Balocco, S., Diaz, O.: The MAMA-SYNTH challenge: Synthesizing virtual contrast-enhancement in breast MRI (2026).https://doi.org/10.5281/ zenodo.19852228

    work page 2026

  14. [14]

    Journal of Medical Imaging12(S2), S22014 (2025)

    Osuala, R., Joshi, S., Tsirikoglou, A., Garrucho, L., Pinaya, W.H.L., Lang, D.M., Schnabel, J.A., Diaz, O., Lekadir, K.: Simulating dynamic tumor contrast enhance- ment in breast MRI using conditional generative adversarial networks. Journal of Medical Imaging12(S2), S22014 (2025). https://doi.org/10.1117/1.JMI.12.S2. S22014

    work page doi:10.1117/1.jmi.12.s2 2025

  15. [15]

    In: Medical Image Computing and Computer Assisted Intervention – MICCAI 2024

    Osuala,R.,Lang,D.M.,Verma,P.,Joshi,S.,Tsirikoglou,A.,Skorupko,G.,Kushibar, K., Garrucho, L., Pinaya, W.H.L., Diaz, O., Schnabel, J.A., Lekadir, K.: Towards learning contrast kinetics with multi-condition latent diffusion models. In: Medical Image Computing and Computer Assisted Intervention – MICCAI 2024. Lecture Notes in Computer Science, vol. 15005, pp....

    work page doi:10.1007/978-3-031-72086-4_67 2024

  16. [16]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Peebles, W., Xie, S.: Scalable diffusion models with transformers. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 4195–4205 (2023).https://doi.org/10.1109/ICCV51070.2023.00387

    arXiv 2023

  17. [17]

    In: Medical Imaging 2019: Image Processing

    Reinhold, J.C., Dewey, B.E., Carass, A., Prince, J.L.: Evaluating the impact of intensity normalization on MR image synthesis. In: Medical Imaging 2019: Image Processing. Proceedings of SPIE, vol. 10949, p. 109493H. SPIE (2019). https://doi.org/10.1117/12.2513089

    work page doi:10.1117/12.2513089 2019

  18. [18]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 10684–10695 (2022). https://doi.org/10.1109/CVPR52688.2022.01042

    arXiv 2022

  19. [19]

    Varma, M., Kumar, A., van der Sluijs, R., Ostmeier, S., Blankemeier, L., Chambon, P., Bluethgen, C., Prince, J., Langlotz, C., Chaudhari, A.: MedVAE: Efficient auto- mated interpretation of medical images with large-scale generalizable autoencoders (2025),https://arxiv.org/abs/2502.14753

    Pith/arXiv arXiv 2025

  20. [20]

    Radiology298(1), 28–35 (2021).https://doi.org/10.1148/radiol.2020202903

    Weinreb,J.C.,Rodby,R.A.,Yee,J.,Wang,C.L.,Fine,D.,McDonald,R.J.,Perazella, M.A., Dillman, J.R., Davenport, M.S.: Use of intravenous gadolinium-based contrast media in patients with kidney disease: Consensus statements from the american college of radiology and the national kidney foundation. Radiology298(1), 28–35 (2021).https://doi.org/10.1148/radiol.2020202903

    work page doi:10.1148/radiol.2020202903 2021

  21. [21]

    https://doi.org/10.5281/ zenodo.8068383

    Zhang, J.: Breast cancer DCE-MRI data (2023). https://doi.org/10.5281/ zenodo.8068383

    work page 2023

  22. [22]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Zhang, Y., Yuan, Y., Song, Y., Wang, H., Liu, J.: EasyControl: Adding efficient and flexible control for diffusion transformer. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 19513–19524 (2025).https: //doi.org/10.1109/ICCV51701.2025.01814

    arXiv 2025

  23. [23]

    Radiology 295(2), 328–338 (2020).https://doi.org/10.1148/radiol.2020191145

    Zwanenburg, A., Vallières, M., Abdalah, M.A., Aerts, H.J.W.L., Andrearczyk, V., Apte, A.P., et al.: The image biomarker standardization initiative: Standardized quantitative radiomics for high-throughput image-based phenotyping. Radiology 295(2), 328–338 (2020).https://doi.org/10.1148/radiol.2020191145

    doi:10.1148/radiol.2020191145 2020

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.