Pith. sign in

REVIEW 4 major objections 4 minor 43 references

Conformalized quantile regression produces pixelwise uncertainty intervals for accelerated MRI reconstruction that track the true reconstruction error with roughly 90% linear correlation at 4x acceleration and above, without any ground-trut

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 ↗

Conformalized quantile regression gives pixel-wise uncertainty intervals for accelerated MRI reconstructions that closely track true reconstruction error, with brain test correlations near or above 0.90 at 4–10x acceleration.

T0 review reviewed 2026-08-03 challenge →

load-bearing objection Useful pixelwise conformal UQ for accelerated MRI, but the headline ~90% correlation likely leans on background pixels; needs a masked re-analysis before the central claim is trusted. the 4 major comments →

arxiv 2601.13236 v3 pith:TAWFEGUC submitted 2026-01-19 eess.IV cs.AIphysics.med-ph

Pixelwise Uncertainty Quantification of Accelerated MRI Reconstruction

Ilias I. Giannakopoulos , Lokesh B Gautham Muthukumar , Yvonne W. Lui , Riccardo Lattanzi This is my paper
classification eess.IV cs.AIphysics.med-ph
keywords conformal predictionquantile regressionuncertainty quantificationaccelerated MRIparallel imagingpixelwise confidence intervalsreconstruction erroranomaly detection
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

The paper tries to establish that pixelwise uncertainty intervals computed by conformalized quantile regression, wrapped around an existing accelerated MRI reconstruction network, can serve as a faithful surrogate for the true reconstruction error even when fully sampled ground-truth images are unavailable. Across brain and knee scans at acceleration factors from 2 to 10, the calibrated interval width correlates with the actual error at a linear correlation above 90% for brain images at 4x and higher, whereas a simpler residual-magnitude heuristic falls below 70% (and below 25% for knees with a small calibration set). If correct, this gives clinicians and automated protocols a per-pixel confidence signal that flags unreliable regions, lesions, and artifacts without slowing the scan or needing a reference image. The framework is an add-on to any reconstruction network and carries finite-sample coverage guarantees from conformal calibration.

Core claim

The paper claims that pixelwise uncertainty intervals produced by conformalized quantile regression—two U-Nets trained with the pinball loss to predict lower and upper conditional quantiles of the reconstruction, followed by conformal calibration of a single scaling factor on a holdout set—yield a statistically valid and spatially informative uncertainty map for accelerated MRI reconstruction. After calibration, the width of these intervals correlates with the true reconstruction error at a linear correlation above 90% for brain images at 4x acceleration and higher, while the residual-magnitude heuristic stays below 70% (and below 25% for knee images with a smaller calibration set). The unce

What carries the argument

The engine is conformalized quantile regression (CQR). Two U-Nets produce pixelwise lower and upper quantile estimates of the reconstruction; they are trained with the pinball loss against ground-truth images. A conformal calibration step then finds the smallest global scaling factor lambda that rescales the quantile offsets so that, on a held-out calibration set, the fraction of pixels whose ground truth falls outside the interval is controlled at the target level (alpha = 0.1), using a conservative finite-sample bound on the coverage. The calibrated interval width is the uncertainty map. CQR is what converts heuristic network outputs into intervals with a statistical guarantee.

Load-bearing premise

The whole guarantee rests on the calibration set being large enough and representative of the scans the method will later see; if the scanner, patient population, or undersampling pattern shifts, the calibrated intervals and the ~90% correlation need not transfer.

What would settle it

Take the calibrated module and evaluate it on brain or knee images from a different scanner or with a pathology distribution not represented in the calibration set; if the linear correlation between interval width and true reconstruction error falls to the level of the residual-magnitude baseline (below 70%, or below 25% with a small calibration set), the central claim that QR uncertainty is a faithful, portable surrogate for reconstruction error is refuted. A simpler internal check: use only 50 calibration volumes on brain data and see if the correlation collapses as it does for the knee.

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

If this is right

  • At 4x acceleration and above, the calibrated interval width acts as a per-pixel proxy for reconstruction error without needing a ground-truth reference at inference.
  • The method transfers across brain and knee anatomies, multiple contrasts, and acceleration factors up to 10x, and the uncertainty maps spatially align with lesions and artifacts.
  • Regions where uncertainty exceeds a threshold derived from the 4x-acceleration maximum could flag slices that should not be read diagnostically or that need re-acquisition.
  • Because the uncertainty module is an add-on trained with the reconstruction network frozen, it can be attached to any reconstruction method without retraining the base model.
  • The framework enables a fast, highly accelerated pre-scan to act as an anomaly detector, potentially deciding when a longer diagnostic-quality scan is necessary.

Where Pith is reading between the lines

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

  • If the ~90% correlation persists across scanner vendors and pathology mixes not seen in training, the approach could let a radiologist set a per-scan 'trust threshold' and automatically flag slices for re-acquisition; the paper does not test this transfer.
  • The single global scaling factor lambda rescales all pixels uniformly; a spatially adaptive calibration (per-slice or per-region lambda) might sharpen the already strong local correlations the paper reports in region-based metrics, an extension the paper leaves open.
  • A concrete testable extension is to run a 10x pre-scan, measure the volume of high-uncertainty regions, and compare it with the lesion volume from a 4x diagnostic scan to quantify sensitivity and specificity as an anomaly detector.
  • Because the uncertainty module is trained after the reconstruction network is frozen, jointly fine-tuning the reconstruction network to reduce high-uncertainty pixels could turn the uncertainty map from a passive indicator into an active training signal; the paper lists this as future work.
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

4 major / 4 minor

Summary. The paper proposes a conformal quantile regression framework for pixelwise uncertainty quantification in accelerated MRI reconstruction. Two U-Nets parameterize lower and upper quantile bounds of the ground-truth image conditioned on the output of a fixed E2E VarNet reconstruction; the bounds are trained with the pinball loss, and a scalar conformal factor is calibrated on a held-out split. The method is evaluated on fastMRI brain and knee data at accelerations 2x to 10x, with the primary quantitative claim being high Pearson correlation (roughly 90% for brain at 4x and above) between the calibrated interval width and the true absolute reconstruction error, plus qualitative evidence that uncertainty highlights lesions and artifacts.

Significance. If the quantitative claims are established, the framework addresses a clinically important problem: providing per-pixel confidence estimates for accelerated MRI reconstructions without access to a fully sampled reference at inference. The study has methodological strengths: separate training/validation/calibration/test splits, comparison against a residual-magnitude baseline, evaluation across multiple acceleration factors and two anatomies, and qualitative lesion-localization evidence. The use of conformal calibration with a concentration bound is appropriate in spirit. However, the current manuscript does not report empirical test-set coverage, the correlation metric is computed on unmasked, Gaussian-blurred images, and the definition of the calibration target is internally inconsistent. These issues must be resolved before the central claims can be accepted.

major comments (4)
  1. [Section II-D2 and II-D4] The notation for α is internally inconsistent. The text defines the upper/lower quantiles as (1−α/2) and α/2, which is standard when α is the miscoverage rate (e.g., α=0.1 for 90% coverage). However, the text then states 'We set α = 90%', and in II-D4 the calibration target is that the average fraction of pixels outside the interval 'does not exceed α'. If α=0.9, the trained interval is only the central 10% interval and the calibration would allow 90% miscoverage. If the intended coverage is 90%, the value of α and the formulas must be corrected (e.g., α=0.1, upper=1−α/2, lower=α/2). As written, the statistical target of the method is not well defined.
  2. [Section II-D4 / Table II] No empirical coverage is reported on the test set. The paper claims conformal calibration yields statistically rigorous, finite-sample valid pixelwise intervals, and λ is chosen using a Hoeffding bound on a calibration set, but nowhere is the achieved coverage (e.g., the fraction of ground-truth pixels inside T_λ) reported for the held-out test data. Since coverage is the primary statistical guarantee, a new table should report pooled and per-image empirical coverage for each acceleration, anatomy, and method, ideally stratified by foreground/background. Without this, the 'statistically rigorous' claim is unsupported.
  3. [Section II-E, Eq. (6), Table II] The headline correlations may be inflated by background pixels. Pearson/Spearman correlations are computed over 'all pixel values within the image' after a Gaussian blur with σ=2; no foreground or tissue mask is used. FastMRI brain and knee images contain large regions of near-zero background where both the reconstruction error and the predicted uncertainty are close to zero, producing a dominant cluster at (0,0) that can raise Pearson (and Spearman) correlations even if uncertainty tracks error poorly inside the anatomy. The region-based correlations are also not masked. The authors should recompute correlations using a foreground/anatomy mask and, as a sensitivity check, without Gaussian blur, and report whether the ~90% claim survives.
  4. [Section II-D4] The finite-sample guarantee relies on a Hoeffding bound applied to 'the average fraction of all pixels' in the calibration set. Pixel residuals are not independent: they are spatially correlated within an image and across slices. If the bound is computed over pooled pixels, plain Hoeffding's inequality is not valid. The risk definition, the unit of exchangeability (pixels vs. images), and the concentration inequality need to be specified precisely. A valid approach would define a per-image risk and apply the bound over the I calibration images, or use a dependence-aware bound. As written, the calibration step may not provide the claimed distribution-free coverage guarantee.
minor comments (4)
  1. [Abstract] The statement that 'Pearson correlation coefficient was higher than 90%' applies only to brain; Table II shows knee Pearson correlations of 0.80±0.16 and 0.82±0.15. Please qualify this claim as brain-specific in the abstract.
  2. [Section II-D2/D4] The term 'pixelwise' should be used carefully. The scalar conformal factor provides marginal coverage over the pooled pixel population, not per-pixel conditional coverage. The authors should clarify this distinction in the text to avoid overclaiming.
  3. [Section II-E] For the region-based correlations, specify how patches that are entirely background or constant are handled, and state whether ties in Spearman rank computation (many zero-valued pixels) are treated consistently.
  4. [Table I] Clarify which method the 'ResM (%)' columns refer to and why ResM is only reported at 4x; also explicitly define the normalization by maximum reconstruction magnitude in the table caption.

Circularity Check

0 steps flagged

No circular derivation: supervised quantile training, coverage-only calibration, and held-out evaluation are not tautological.

full rationale

The claimed derivation is not circular. The two quantile U-Nets are trained with the pinball loss (Eqs. 1-2) on (x,y) pairs, but the reported quantity of interest, the calibrated interval width q = ub - lb (Eqs. 4,6), is not defined in terms of the error e = |x-y|, nor is the conformal factor lambda (Eq. 5) chosen to maximize correlation—it is chosen only to control average coverage. Evaluation then measures association between q and e on a held-out test set (Eqs. 7-8). Supervised training on ground truth is standard for a predictive model and does not make the held-out correlation a tautology; the QR loss targets conditional quantiles, not the error map, and the ResM baseline that explicitly regresses e (Eq. 3) makes the comparison meaningful rather than circular. The self-citation [12] is contextual clinical background, not load-bearing for the uncertainty derivation. The Discussion's own caveat that calibration requires a 'relatively large and in-distribution held-out calibration set' is a transferability limitation, and the absence of an explicit background mask in the correlation analysis is a potential confound, but neither reduces the central claim to its inputs by construction.

Axiom & Free-Parameter Ledger

2 free parameters · 6 axioms · 0 invented entities

The core method relies on standard conformal prediction theory and supervised quantile-regression training; the paper's own contribution is the architecture and calibration adaptation. The main postulates are exchangeability and accuracy of fastMRI ground truth; the anomaly-detection interpretation additionally assumes reconstruction errors concentrate on pathologies.

free parameters (2)
  • λ (conformal scaling factor per acceleration/anatomy/model) = brain QR: 1.31, 1.54, 1.59, 1.74, 1.87 (2×–10×); brain ResM 4×: 1.02; knee QR: 1.42 (4×), 1.68 (6×); knee ResM 4×: 0.82
    Chosen on the calibration set by minimizing λ such that the conservative Hoeffding bound on pixel miscoverage is ≤ α; directly controls interval width and thus the reported uncertainty values.
  • Gaussian blur σ for correlation evaluation = 2 pixels
    Chosen by hand to 'suppress pixel-level noise' (Sec II-E); applied to both predicted uncertainty and true error before Pearson/Spearman correlations, so it can inflate reported agreement.
axioms (6)
  • domain assumption Train/calibration/test data are exchangeable (same fastMRI distribution, random split)
    The conformal guarantee and calibration factor λ rely on this; the paper splits validation randomly into 689/689 volumes and uses a fixed test set (Sec II-A).
  • domain assumption Fully sampled fastMRI ground-truth references are treated as accurate y
    Used to train quantiles and to compute error maps; any noise in references propagates into both terms and can affect correlations.
  • standard math Quantile regression U-Nets converge to true conditional quantiles
    The paper relies on asymptotic consistency of pinball-loss minimizers (cites Koenker-Bassett, Romano et al.), but no finite-sample convergence result is proven for this architecture.
  • ad hoc to paper Reconstruction error is larger at pathologies (used for anomaly-detection interpretation)
    Introduced in Discussion ('If we assume...'), not quantitatively validated with lesion segmentations; drives the qualitative claim that uncertainty can detect anomalies.
  • standard math Hoeffding inequality bound for conservative coverage calibration
    Used to choose λ with finite-sample coverage (Sec II-D4).
  • domain assumption The trained E2E VarNet implementation is a faithful instance of the published model
    The paper re-trains VarNet with stated hyperparameters; SSIM values in Table I are plausible but no model/code is released.

reviewed 2026-08-03 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Pixelwise Uncertainty Quantification of Accelerated MRI Reconstruction." pith.science (2026). https://pith.science/paper/TAWFEGUC

@misc{pith2026260113236,
  author       = {Pith},
  title        = {Pith review of: Pixelwise Uncertainty Quantification of Accelerated MRI Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TAWFEGUC}},
  note         = {Machine review of arXiv:2601.13236}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Parallel imaging techniques reduce magnetic resonance imaging (MRI) scan time but image quality degrades as the acceleration factor increases. In clinical practice, conservative acceleration factors are chosen because no mechanism exists to automatically assess the diagnostic quality of undersampled reconstructions. This work introduces a general framework for pixel-wise uncertainty quantification in parallel MRI reconstructions, enabling automatic identification of unreliable regions without access to any ground-truth reference image. Our method integrates conformal quantile regression with image reconstruction methods to estimate statistically rigorous pixel-wise uncertainty intervals. We trained and evaluated our model on Cartesian undersampled brain and knee data obtained from the fastMRI dataset using acceleration factors ranging from 2 to 10. An end-to-end Variational Network was used for image reconstruction. Quantitative experiments demonstrate strong agreement between predicted uncertainty maps and true reconstruction error. Using our method, the corresponding Pearson correlation coefficient was higher than 90% at acceleration levels at and above four-fold; whereas it dropped to less than 70% when the uncertainty was computed using a simpler a heuristic notion (magnitude of the residual). Qualitative examples further show the uncertainty maps based on quantile regression capture the magnitude and spatial distribution of reconstruction errors across acceleration factors, with regions of elevated uncertainty aligning with pathologies and artifacts. The proposed framework enables evaluation of reconstruction quality without access to fully-sampled ground-truth reference images. It represents a step toward adaptive MRI acquisition protocols that may be able to dynamically balance scan time and diagnostic reliability.

Figures

Figures reproduced from arXiv: 2601.13236 by Ilias I. Giannakopoulos, Lokesh B Gautham Muthukumar, Riccardo Lattanzi, Yvonne W. Lui.

Figure 1
Figure 1. Figure 1: E2E VarNet architecture. The input is the undersampled multi-coil k-space and the output is the reconstructed image. In each cascade, the inverse fast Fourier transform (iFFT) is applied to each undersampled k-space, and the resulting images are weighted with the corresponding coil sensitivities and are combined into a single image using the reduce operator [1]. The combined image is processed by a U-Net, … view at source ↗
Figure 2
Figure 2. Figure 2: (Left) Architecture of our proposed uncertainty estimation module. The output (reconstructed MR image) of the E2E VarNet is the input of two U-Nets, which learn pixelwise offsets that parameterize the lower and upper quantile bounds of the reconstruction uncertainty. Each U-Net output passes through a sigmoid activation, is multiplied by the E2E VarNet reconstruction, and is either added to or subtracted f… view at source ↗
Figure 3
Figure 3. Figure 3: shows the distribution of the Pearson correlation for all test data between the predicted uncertainty and the true reconstruction error for the 4× undersampled brain (top) and knee (bottom) test dataset. Each bar represents the number of cases achieving a given correlation level. The QR-based estimates (blue) are consistently higher than the ResM-based ones (red) for both anatomies [PITH_FULL_IMAGE:figure… view at source ↗
Figure 4
Figure 4. Figure 4: Comparison between the windowed absolute error (magnified 50 times), the QR-based, and the ResM-based uncertainty for one healthy and three abnormal brain cases. All reconstructions were performed with four-fold acceleration. Blurred versions of the error and the uncertainties, generated by Gaussian smoothing, are also shown. The QR-based uncertainty closely matches the absolute error distribution and in t… view at source ↗
Figure 5
Figure 5. Figure 5: Comparison between the windowed absolute error (magnified 50 times), the QR-based, and the ResM-based uncertainty for three knees. All reconstructions were performed with four-fold acceleration. Blurred versions of the error and the uncertainties, generated by Gaussian smoothing, are also shown. The QR-based uncertainty qualitatively matches the absolute error distribution unlike the ResM-based approach [… view at source ↗
Figure 6
Figure 6. Figure 6: Reconstructed images and corresponding QR-based uncertainty maps for one abnormal brain for five acceleration factors. Starting at four-fold acceleration, the QR-based uncertainty increases at the location of the lesion and reflects the posterior susceptibility artifact from a prior craniotomy. The bottom row overlays the uncertainty heatmaps on the corresponding reconstructed images to better illustrate t… view at source ↗
Figure 7
Figure 7. Figure 7: QR-based uncertainty maps for a healthy brain and three abnormal brain cases across increasing acceleration factors. For each case, uncertainty maps are shown for 2, 4, 6, 8, and 10× acceleration to illustrate how uncertainty evolves with undersampling. Top: unthresholded uncertainty maps (lower bound set to zero). Bottom: uncertainty maps thresholded using the maximum uncertainty observed at 4× accelerati… view at source ↗
Figure 8
Figure 8. Figure 8: QR-based uncertainty maps for three knee cases at increasing acceleration. For each case, uncertainty maps are shown for 4 and 6× acceleration to illustrate how uncertainty changes with undersampling in knee MRI. Top: unthresholded uncertainty maps (lower bound set to zero). Bottom: uncertainty maps thresholded using the maximum uncertainty observed at 4× acceleration. Thresholding highlights regions where… 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

43 extracted references · 3 linked inside Pith

  1. [1]

    End-to-end variational networks for accelerated MRI reconstruction,

    A. Sriram et al. , “End-to-end variational networks for accelerated MRI reconstruction,” in International conference on medical image computing and computer-assisted intervention . Springer, 2020, pp. 64–73

    2020

  2. [2]

    Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays,

    D. K. Sodickson and W. J. Manning, “Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays,” Magnetic resonance in medicine , vol. 38, no. 4, pp. 591–603, 1997

    1997

  3. [3]

    SENSE: sensitivity encoding for fast MRI,

    K. P . Pruessmann, M. Weiger, M. B. Scheidegger, and P . Boesiger, “SENSE: sensitivity encoding for fast MRI,” Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine , vol. 42, no. 5, pp. 952–962, 1999

    1999

  4. [4]

    Generalized autocalibrating partially parallel acquisitions (GRAPPA),

    M. A. Griswold et al. , “Generalized autocalibrating partially parallel acquisitions (GRAPPA),” Magnetic Resonance in Medicine: An Offi- cial Journal of the International Society for Magnetic Resonance in Medicine, vol. 47, no. 6, pp. 1202–1210, 2002

    2002

  5. [5]

    Sparse MRI: The application of compressed sensing for rapid MR imaging,

    M. Lustig, D. Donoho, and J. M. Pauly, “Sparse MRI: The application of compressed sensing for rapid MR imaging,” Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine , vol. 58, no. 6, pp. 1182–1195, 2007

    2007

  6. [6]

    Image reconstruction by regularized nonlinear inversionjoint estimation of coil sensitivities and image content,

    M. Uecker, T. Hohage, K. T. Block, and J. Frahm, “Image reconstruction by regularized nonlinear inversionjoint estimation of coil sensitivities and image content,” Magnetic Resonance in Medicine , vol. 60, no. 3, pp. 674–682, 2008

    2008

  7. [7]

    Learning a variational network for reconstruction of accelerated MRI data,

    K. Hammernik et al., “Learning a variational network for reconstruction of accelerated MRI data,” Magnetic resonance in medicine , vol. 79, no. 6, pp. 3055–3071, 2018

    2018

  8. [8]

    MoDL: Model-based deep learning architecture for inverse problems,

    H. K. Aggarwal, M. P . Mani, and M. Jacob, “MoDL: Model-based deep learning architecture for inverse problems,” IEEE transactions on medical imaging , vol. 38, no. 2, pp. 394–405, 2018

    2018

  9. [9]

    KIKI-net: cross-domain convolutional neural networks for reconstructing under- sampled magnetic resonance images,

    T. Eo, Y . Jun, T. Kim, J. Jang, H.-J. Lee, and D. Hwang, “KIKI-net: cross-domain convolutional neural networks for reconstructing under- sampled magnetic resonance images,” Magnetic resonance in medicine , vol. 80, no. 5, pp. 2188–2201, 2018

    2018

  10. [10]

    Deep J-Sense: Accelerated MRI reconstruction via unrolled alternating optimization,

    M. Arvinte, S. Vishwanath, A. H. Tewfik, and J. I. Tamir, “Deep J-Sense: Accelerated MRI reconstruction via unrolled alternating optimization,” in International conference on medical image computing and computer- assisted intervention . Springer, 2021, pp. 350–360

    2021

  11. [11]

    Joint deep model-based MR image and coil sensitivity reconstruction network (joint-ICNet) for fast MRI,

    Y . Jun, H. Shin, T. Eo, and D. Hwang, “Joint deep model-based MR image and coil sensitivity reconstruction network (joint-ICNet) for fast MRI,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2021, pp. 5270–5279

    2021

  12. [12]

    Accelerated MRI reconstructions via varia- tional network and feature domain learning,

    I. I. Giannakopoulos et al. , “Accelerated MRI reconstructions via varia- tional network and feature domain learning,” Scientific Reports , vol. 14, no. 1, p. 10991, 2024

    2024

  13. [13]

    Using deep learning to accelerate knee MRI at 3 T: results of an interchangeability study,

    M. P . Recht et al. , “Using deep learning to accelerate knee MRI at 3 T: results of an interchangeability study,” American Journal of Roentgenology, vol. 215, no. 6, pp. 1421–1429, 2020

    2020

  14. [14]

    Artificial intelligence for MR image reconstruction: an overview for clinicians,

    D. J. Lin, P . M. Johnson, F. Knoll, and Y . W. Lui, “Artificial intelligence for MR image reconstruction: an overview for clinicians,” Journal of Magnetic Resonance Imaging , vol. 53, no. 4, pp. 1015–1028, 2021

    2021

  15. [15]

    Deep learning reconstruction enables prospectively accelerated clinical knee MRI,

    P . M. Johnsonet al., “Deep learning reconstruction enables prospectively accelerated clinical knee MRI,” Radiology, vol. 307, no. 2, p. e220425, 2023

    2023

  16. [16]

    Exploring the acceleration limits of deep learning variational network–based two-dimensional brain MRI,

    A. Radmanesh et al., “Exploring the acceleration limits of deep learning variational network–based two-dimensional brain MRI,” Radiology: Artificial Intelligence , vol. 4, no. 6, p. e210313, 2022

    2022

  17. [17]

    Prostate Cancer Risk Stratification and Scan Tailoring Using Deep Learning on Abbreviated Prostate MRI,

    P . M. Johnson et al. , “Prostate Cancer Risk Stratification and Scan Tailoring Using Deep Learning on Abbreviated Prostate MRI,” Journal of Magnetic Resonance Imaging , 2025

    2025

  18. [18]

    On hallucinations in tomographic image reconstruction,

    S. Bhadra, V . A. Kelkar, F. J. Brooks, and M. A. Anastasio, “On hallucinations in tomographic image reconstruction,” IEEE transactions on medical imaging , vol. 40, no. 11, pp. 3249–3260, 2021

    2021

  19. [19]

    Clinical impact of deep learning reconstruction in MRI,

    S. Kiryu et al., “Clinical impact of deep learning reconstruction in MRI,” Radiographics, vol. 43, no. 6, p. e220133, 2023

    2023

  20. [20]

    What uncertainties do we need in bayesian deep learning for computer vision?

    A. Kendall and Y . Gal, “What uncertainties do we need in bayesian deep learning for computer vision?” Advances in neural information processing systems , vol. 30, 2017

    2017

  21. [21]

    A review of uncertainty quantification in deep learning: Techniques, applications and challenges,

    M. Abdar et al., “A review of uncertainty quantification in deep learning: Techniques, applications and challenges,” Information fusion , vol. 76, pp. 243–297, 2021

    2021

  22. [22]

    Image-to-image regression with distribution- free uncertainty quantification and applications in imaging,

    A. N. Angelopoulos et al., “Image-to-image regression with distribution- free uncertainty quantification and applications in imaging,” in Interna- tional Conference on Machine Learning . PMLR, 2022, pp. 717–730

    2022

  23. [23]

    Uncertainty quantification in deep MRI reconstruction,

    V . Edupuganti, M. Mardani, S. V asanawala, and J. Pauly, “Uncertainty quantification in deep MRI reconstruction,” IEEE Transactions on Med- ical Imaging , vol. 40, no. 1, pp. 239–250, 2020

    2020

  24. [24]

    Uncertainty quantification in deep learning for safer neuroimage enhancement,

    R. Tanno et al. , “Uncertainty quantification in deep learning for safer neuroimage enhancement,” arXiv preprint arXiv:1907.13418 , 2019

    Pith/arXiv arXiv 1907

  25. [25]

    Bayesian uncertainty estimation of learned variational MRI reconstruction,

    D. Narnhofer, A. Effland, E. Kobler, K. Hammernik, F. Knoll, and T. Pock, “Bayesian uncertainty estimation of learned variational MRI reconstruction,” IEEE transactions on medical imaging , vol. 41, no. 2, pp. 279–291, 2021

    2021

  26. [26]

    PixCUE: joint uncertainty estimation and image reconstruction in MRI using deep pixel classification,

    M. Ekanayake, K. Pawar, Z. Chen, G. Egan, and Z. Chen, “PixCUE: joint uncertainty estimation and image reconstruction in MRI using deep pixel classification,” Journal of Imaging Informatics in Medicine , vol. 38, no. 4, pp. 2071–2084, 2025

    2071

  27. [27]

    Rigorous uncertainty estimation for MRI reconstruc- tion,

    K. Wang et al. , “Rigorous uncertainty estimation for MRI reconstruc- tion,” in Proceedings of the proceedings of the 30th annual meeting of ISMRM, vol. 749, 2022

    2022

  28. [28]

    CUTE- MRI: Conformalized Uncertainty-based framework for Time-adaptivE MRI,

    P . Fischer, J. N. Morshuis, T. Küstner, and C. Baumgartner, “CUTE- MRI: Conformalized Uncertainty-based framework for Time-adaptivE MRI,” arXiv preprint arXiv:2508.14952 , 2025

    Pith/arXiv arXiv 2025

  29. [29]

    fastMRI: An open dataset and benchmarks for accelerated MRI,

    J. Zbontar et al. , “fastMRI: An open dataset and benchmarks for accelerated MRI,” arXiv preprint arXiv:1811.08839 , 2018

    Pith/arXiv arXiv 2018

  30. [30]

    fastMRI: A publicly available raw k-space and DICOM dataset of knee images for accelerated MR image reconstruction using machine learning,

    F. Knoll et al., “fastMRI: A publicly available raw k-space and DICOM dataset of knee images for accelerated MR image reconstruction using machine learning,” Radiology: Artificial Intelligence , vol. 2, no. 1, p. e190007, 2020

    2020

  31. [31]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P . Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in International Conference on Medical image computing and computer-assisted intervention . Springer, 2015, pp. 234–241

    2015

  32. [32]

    Multiscale structural similarity for image quality assessment,

    Z. Wang, E. P . Simoncelli, and A. C. Bovik, “Multiscale structural similarity for image quality assessment,” in The thrity-seventh asilomar conference on signals, systems & computers, 2003 , vol. 2. Ieee, 2003, pp. 1398–1402

    2003

  33. [33]

    Regression quantiles,

    R. Koenker and G. Bassett Jr, “Regression quantiles,” Econometrica: journal of the Econometric Society , pp. 33–50, 1978

    1978

  34. [34]

    Conformalized quantile regression,

    Y . Romano, E. Patterson, and E. Candes, “Conformalized quantile regression,” Advances in neural information processing systems , vol. 32, 2019

    2019

  35. [35]

    Global nonparametric estimation of conditional quantile functions and their derivatives,

    P . Chaudhuri, “Global nonparametric estimation of conditional quantile functions and their derivatives,” Journal of multivariate analysis , vol. 39, no. 2, pp. 246–269, 1991. 10

    1991

  36. [36]

    Direct use of regression quantiles to construct confidence sets in linear models,

    K. Q. Zhou and S. L. Portnoy, “Direct use of regression quantiles to construct confidence sets in linear models,” The Annals of Statistics , vol. 24, no. 1, pp. 287–306, 1996

    1996

  37. [37]

    Statistical inference on heteroscedas- tic models based on regression quantiles,

    K. Q. Zhou and S. L. Portnoy, “Statistical inference on heteroscedas- tic models based on regression quantiles,” Journal of Nonparametric Statistics, vol. 9, no. 3, pp. 239–260, 1998

    1998

  38. [38]

    Nonparametric quantile estimation,

    I. Takeuchi, Q. Le, T. Sears, A. Smola et al. , “Nonparametric quantile estimation,” Journal of Machine Learning Research , vol. 7, pp. 1231– 1264, 2006

    2006

  39. [39]

    Estimating conditional quantiles with the help of the pinball loss,

    I. Steinwart and A. Christmann, “Estimating conditional quantiles with the help of the pinball loss,” Bernoulli, vol. 17, no. 1, pp. 211–225, 2011

    2011

  40. [40]

    Distribution-free, risk-controlling prediction sets,

    S. Bates, A. Angelopoulos, L. Lei, J. Malik, and M. Jordan, “Distribution-free, risk-controlling prediction sets,” Journal of the ACM (JACM), vol. 68, no. 6, pp. 1–34, 2021

    2021

  41. [41]

    Probability inequalities for sums of bounded random variables,

    W. Hoeffding, “Probability inequalities for sums of bounded random variables,” Journal of the American statistical association , vol. 58, no. 301, pp. 13–30, 1963

    1963

  42. [42]

    VII. Note on regression and inheritance in the case of two parents,

    K. Pearson, “VII. Note on regression and inheritance in the case of two parents,” proceedings of the royal society of London , vol. 58, no. 347-352, pp. 240–242, 1895

  43. [43]

    “General Intelligence

    C. Spearman, ““General Intelligence” Objectively Determined and Mea- sured.” The American Journal of Psychology , vol. 15, no. 2, pp. 201– 293, 1904

    1904

This paper was first reviewed by deepseek-v4-flash on August 3, 2026.