pith. machine review for the scientific record. sign in

arxiv: 2605.09590 · v1 · submitted 2026-05-10 · 📡 eess.SP

Recognition: 2 theorem links

· Lean Theorem

Fast Voxelwise SNR Estimation for Iterative MRI Reconstructions

Onat Dalmaz , Daniel Abraham , Alexander R. Toews , Akshay S. Chaudhari , Kawin Setsompop , Brian A. Hargreaves
Authors on Pith no claims yet

Pith reviewed 2026-05-12 03:45 UTC · model grok-4.3

classification 📡 eess.SP
keywords MRInoise estimationSNR mappingg-factoriterative reconstructioncompressed sensingvoxelwise analysis
0
0 comments X

The pith

PICO recovers voxelwise noise variance in general MRI reconstructions by probing the image-domain covariance operator with random-phase vectors.

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

The paper introduces PICO as a method to estimate noise variance quickly in both linear and nonlinear iterative MRI reconstructions. It does this by applying a few random complex-phase probe images to the image-domain noise covariance operator or its Jacobian. This approach avoids the computational burden of analytical formulas for non-Cartesian sampling and the long runtimes of methods that use many replica reconstructions. Validation on Cartesian, spiral, and compressed-sensing datasets shows it matches reference methods while being several times faster. As a result, detailed SNR and g-factor maps can become a standard output of the reconstruction process.

Core claim

PICO estimates the image-domain noise variance by probing the noise covariance operator with complex random-phase vectors. These probes are shown to minimize estimator variance compared to Gaussian or real-valued alternatives. For nonlinear reconstructions, the Jacobian of the converged solution is probed instead. This yields accurate voxelwise SNR, g-factor, and related metrics without requiring closed-form expressions or large numbers of replica images.

What carries the argument

The probing of the image-domain noise covariance operator (or Jacobian) using complex random-phase probe images, which reuses existing reconstruction primitives to compute the variance estimates efficiently.

If this is right

  • In Cartesian SENSE reconstructions, PICO reproduces analytical g-factor maps accurately.
  • In non-Cartesian spiral imaging at R=2, it achieves 1% error in 64 seconds versus 462 seconds for PMR.
  • For compressed-sensing knee reconstructions, Jacobian probing produces consistent noise maps faster than PMR.
  • The method works across linear and nonlinear iterative reconstructions without additional calibration.

Where Pith is reading between the lines

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

  • Integrating PICO into clinical reconstruction pipelines could make quantitative image quality assessment routine without extra scan time.
  • Similar probing strategies might apply to noise estimation in other iterative reconstruction problems outside MRI, such as CT or ultrasound.
  • Future work could explore optimal probe count or adaptive probing for even lower variance estimates.

Load-bearing premise

The image-domain noise covariance can be sufficiently well approximated by probing it with only a small number of random complex-phase vectors to give unbiased variance estimates.

What would settle it

A direct comparison on a new MRI dataset where PICO noise maps differ significantly from high-replica PMR references or analytical ground truth would show the estimator is inaccurate.

Figures

Figures reproduced from arXiv: 2605.09590 by Akshay S. Chaudhari, Alexander R. Toews, Brian A. Hargreaves, Daniel Abraham, Kawin Setsompop, Onat Dalmaz.

Figure 1
Figure 1. Figure 1: Overview of PICO for fast voxelwise noise characterization. Left: the MRI acquisition is modeled by an encoding operator A (coil sensitivities, sampling trajectory, density compensation); the reconstruction operator R maps multi-coil k-space data to the image estimate xˆ. Right: PICO estimates the voxelwise noise variance map diag(Σxˆ) in four steps. (1) Generate unit-magnitude random-phase probe images v … view at source ↗
Figure 2
Figure 2. Figure 2: Reciprocal g-factor maps (1/g) for Cartesian CG-SENSE at R = 2, shown for increasing probe/replica counts (N = 10, 20, 40, 80, 200). Top: PICO; bottom: PMR. Right column: analytical SENSE reference (top) and reconstructed magnitude image (bottom). All panels share a fixed color scale set by the analytical reference. PICO recovers the spatial structure of the g-factor by N = 10 and is visually converged by … view at source ↗
Figure 3
Figure 3. Figure 3: Voxelwise g-factor maps for non-Cartesian CG-SENSE (R = 2, spiral trajectory). Columns show increasing sample counts (N ∈ {50, 200, 1000, 5000}); right column: PMR reference at N = 30,000 and magnitude image. Top: PICO; bottom: PMR. PICO resolves the g-factor structure by N = 1000, whereas PMR retains visible Monte Carlo artifacts even at N = 5000. N = 9400 replicas (538.6 s; ≈5.9×). Across all acceleratio… view at source ↗
Figure 4
Figure 4. Figure 4: NRMSE vs. runtime for non-Cartesian CG-SENSE (log–log scale). (a) R = 2: PICO reaches 1% NRMSE in 64.0 s (N = 800); PMR requires 462.6 s (N = 7950). (b) R = 3: proposed 70.6 s (N = 900) vs. PMR 501.0 s (N = 8600). (c) R = 4: proposed 91.0 s (N = 1150) vs. PMR 538.6 s (N = 9400). Dashed line: 1% NRMSE threshold; triangles: first crossing for each method [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: NRMSE vs. probe count N for three probe families on non-Cartesian CG-SENSE (log–log scale). Random phase (blue) achieves the lowest error at every N, followed by real Rademacher (green) and complex Gaussian (red), confirming the theoretical optimality of unit-magnitude random-phase probes. All three distributions exhibit approximately parallel power-law decay, with the vertical offset governed by the probe… view at source ↗
Figure 6
Figure 6. Figure 6: Noise standard deviation maps for TV-regularized CS reconstruction (R = 2). Top: zero-filled and CS magnitude images. Bottom: noise maps from PICO (left) and PMR (right), with slice-specific runtimes. The two methods produce visually identical maps; PICO converges faster (e.g., 66 s vs. 122 s for Slice 1). (A, AH, Toeplitz normal operators, inner CG steps), avoiding repeated end-to-end solves. A second adv… view at source ↗
Figure 7
Figure 7. Figure 7: Robustness to input noise level in nonlinear CS reconstruction (N = 3000). Columns sweep the noise standard deviation from σ0 to 200σ0 (SNR ≈ 45 to −1 dB). Top: magnitude images. Middle: normalized noise maps (σout/σk) from PICO and PMR. Bottom: absolute difference and ×10 amplified view. At moderate-to-high SNR (≥ 16 dB), the two methods agree closely. At extreme noise levels (≤ 4 dB), residual discrepanc… view at source ↗
read the original abstract

Purpose: To develop a fast, general-purpose framework for voxelwise noise characterization in linear and nonlinear iterative MRI reconstructions, recovering the image-domain noise variance from which SNR, $g$-factor, and related image-quality metrics are derived. The framework addresses both the intractability of closed-form formulas beyond Cartesian sampling and the long runtime of Pseudo Multiple Replica (PMR) methods. Methods: We propose PICO (Probing Image-space COvariance), an estimator that operates in the image domain by probing the image-domain noise covariance operator -- or, for nonlinear compressed-sensing reconstructions, the Jacobian of the converged solution -- with random probe images. Complex random-phase probes are shown theoretically and empirically to minimize estimator variance compared with Gaussian or real-valued alternatives. PICO was validated against analytical benchmarks and high-replica PMR references using retrospective Cartesian knee data ($R=2$), prospective non-Cartesian spiral brain phantom data ($R=2,3,4$), and compressed-sensing knee reconstructions ($R=2$). Results: In Cartesian experiments, PICO accurately reproduced analytical SENSE $g$-factor maps. In non-Cartesian spiral imaging ($R=2$), it achieved 1% estimation error in 64 s compared with 462 s for PMR (approximately 7.2x speedup), with the efficiency advantage persisting at higher acceleration. For nonlinear compressed sensing, the Jacobian-based estimator produced noise maps consistent with PMR while converging faster (52 s vs. 95 s; approximately 1.8x speedup). Conclusion: PICO provides a computationally efficient alternative to PMR for voxelwise noise and $g$-factor estimation across generalized iterative MRI reconstructions. By reusing existing reconstruction primitives, it enables voxelwise noise maps to be produced as a routine by-product of the reconstruction pipeline.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes PICO (Probing Image-space COvariance), a framework for fast voxelwise noise variance estimation in linear and nonlinear iterative MRI reconstructions. It recovers per-voxel noise statistics by applying a small number of complex random-phase probes to the image-domain noise covariance operator (or its Jacobian at convergence for compressed-sensing cases), derives SNR and g-factor maps from the resulting variance estimates, and reports substantial speedups over Pseudo Multiple Replica (PMR) while matching analytical SENSE references on Cartesian data and PMR on non-Cartesian spiral and CS knee data.

Significance. If the finite-probe estimator is unbiased and its variance remains low enough to preserve the reported accuracy, PICO would enable routine voxelwise noise mapping as a low-overhead byproduct of generalized iterative reconstructions, addressing a practical bottleneck in quantitative MRI. The explicit validation against both analytical g-factor maps and high-replica PMR on multiple sampling schemes, together with the reuse of existing reconstruction operators, strengthens its potential utility.

major comments (2)
  1. [Methods, PICO Estimator] Methods (PICO estimator derivation): the unbiasedness of the diagonal covariance estimate obtained from a modest number of complex random-phase probes is asserted on the basis of covariance-operator properties, yet the explicit expectation calculation showing E[probe estimate] equals the true per-voxel variance for finite probe count (rather than only asymptotically) is not provided; without it the 1 % error figures in the spiral experiments cannot be guaranteed independent of probe count.
  2. [Results, non-Cartesian and CS experiments] Results, non-Cartesian and CS sections: the reported speedups (7.2× for R=2 spiral, 1.8× for CS) and error levels presuppose that estimator variance with the chosen probe count stays below the threshold that would negate the advantage over PMR; no separate analysis or table quantifies the Monte-Carlo variance of the PICO estimator itself across repeated probe realizations.
minor comments (2)
  1. [Abstract and Methods] The abstract states that complex phases are 'theoretically and empirically optimal' for variance reduction; the corresponding variance formula or comparison table should be referenced in the main text for readers who do not consult the supplement.
  2. [Figures] Figure captions for the g-factor and noise maps should explicitly state the number of probes used in each PICO reconstruction so that the timing and accuracy numbers can be reproduced.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review of our manuscript on the PICO framework. We address each major comment point by point below and have revised the manuscript to incorporate the requested clarifications.

read point-by-point responses

  1. Referee: [Methods, PICO Estimator] Methods (PICO estimator derivation): the unbiasedness of the diagonal covariance estimate obtained from a modest number of complex random-phase probes is asserted on the basis of covariance-operator properties, yet the explicit expectation calculation showing E[probe estimate] equals the true per-voxel variance for finite probe count (rather than only asymptotically) is not provided; without it the 1 % error figures in the spiral experiments cannot be guaranteed independent of probe count.

    Authors: We thank the referee for this observation. The manuscript asserts unbiasedness from the fact that complex random-phase probes p satisfy E[p p^H] = I exactly (due to uniform phase distribution over the unit circle), so that the element-wise estimator for the diagonal of the image-domain covariance operator C has expectation exactly equal to diag(C) for any finite N. The average over N probes is therefore an unbiased estimator, with variance decreasing as 1/N. We agree that an explicit derivation would strengthen the presentation and will add it to the Methods section in the revised manuscript, showing E[(C p) ⊙ conj(p)] = diag(C) step by step. This confirms that the reported 1% errors hold for the modest probe counts used and are not reliant on asymptotic arguments. revision: yes

  2. Referee: [Results, non-Cartesian and CS experiments] Results, non-Cartesian and CS sections: the reported speedups (7.2× for R=2 spiral, 1.8× for CS) and error levels presuppose that estimator variance with the chosen probe count stays below the threshold that would negate the advantage over PMR; no separate analysis or table quantifies the Monte-Carlo variance of the PICO estimator itself across repeated probe realizations.

    Authors: We acknowledge that the manuscript does not provide a dedicated quantification of the Monte-Carlo variance of the PICO estimator across independent probe realizations. The 1% error values reflect agreement with high-replica PMR references (whose own variance is negligible), and the speedups are wall-clock comparisons at matched accuracy. To address the referee's concern directly, we will add a brief supplementary analysis (new table or paragraph in the revised Results) reporting the standard deviation of PICO estimates over 20 independent probe realizations for the spiral and CS cases. This will show that estimator variance remains low enough (relative standard deviation well below 1%) to preserve the accuracy and speedup claims. revision: yes

Circularity Check

0 steps flagged

No circularity; derivation from standard random probing of covariance operators

full rationale

The PICO estimator is constructed by applying random complex-phase vectors to the image-domain noise covariance operator (or Jacobian at convergence), with variance estimates obtained via the standard expectation property E[|probe^H * op * probe|] for the diagonal. This follows directly from linearity of expectation and properties of covariance operators without any fitted parameters, self-referential definitions, or load-bearing self-citations that reduce the result to its inputs. Validation against independent analytical SENSE g-factor maps and high-replica PMR references further confirms the chain is externally grounded rather than tautological. No equations in the provided description equate outputs to inputs by construction, and complex-phase optimality is shown via variance minimization rather than ansatz smuggling or renaming.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The framework rests on standard linear algebra for covariance probing and domain assumptions from MRI reconstruction theory; no new entities are postulated and only minor implementation choices (probe count, phase distribution) are introduced without independent evidence.

free parameters (1)
  • number of probes
    The count of random probes used to estimate the covariance; chosen to trade off estimator variance against runtime but not fitted to data in the reported experiments.
axioms (2)
  • domain assumption The reconstruction operator is differentiable or linear so that the Jacobian or covariance operator exists and can be probed.
    Invoked when extending the method to nonlinear compressed-sensing reconstructions via the Jacobian of the converged solution.
  • domain assumption Complex random-phase probes minimize estimator variance relative to Gaussian or real-valued probes.
    Stated as shown theoretically and empirically but the proof is not expanded in the abstract.

pith-pipeline@v0.9.0 · 5654 in / 1390 out tokens · 62993 ms · 2026-05-12T03:45:52.185345+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

73 extracted references · 73 canonical work pages

  1. [1]

    Noise in mri.Magnetic Resonance in Medicine, 36(3):494–497, 1996

    Albert Macovski. Noise in mri.Magnetic Resonance in Medicine, 36(3):494–497, 1996. doi: https://doi.org/10.1002/mrm.1910360327. URL https://onlinelibrary.wiley.com/doi/ abs/10.1002/mrm.1910360327

    work page doi:10.1002/mrm.1910360327 1996

  2. [3]

    Raya, Scott B

    Olaf Dietrich, José G. Raya, Scott B. Reeder, Michael Ingrisch, Maximilian F. Reiser, and Stefan O. Schoenberg. Influence of multichannel combination, parallel imaging and other reconstruction techniques on mri noise characteristics.Magnetic Resonance Imaging, 26 (6):754–762, 2008. ISSN 0730-725X. doi: https://doi.org/10.1016/j.mri.2008.02.001. URL https:...

    work page doi:10.1016/j.mri.2008.02.001 2008

  3. [4]

    Raya, Scott B

    Olaf Dietrich, José G. Raya, Scott B. Reeder, Maximilian F. Reiser, and Stefan O. Schoenberg. Measurement of signal-to-noise ratios in mr images: Influence of multichannel coils, parallel imaging, and reconstruction filters.Journal of Magnetic Resonance Imaging, 26:375–385, 8

    work page

  4. [5]

    doi: 10.1002/jmri.20969

    ISSN 10531807. doi: 10.1002/jmri.20969

    work page doi:10.1002/jmri.20969

  5. [6]

    Peter Kellman and Elliot R. McVeigh. Image reconstruction in snr units: A general method for snr measurement.Magnetic Resonance in Medicine, 54:1439–1447, 2005. ISSN 07403194. doi: 10.1002/mrm.20713

    work page doi:10.1002/mrm.20713 2005

  6. [7]

    Rubenstein, John M

    James D. Rubenstein, John M. Brown, John C. Kohn, and Steven P. Arnoczky. Cartilage invasion by fat: A possible mechanism of steroid-induced osteonecrosis?American Journal 18 of Roentgenology, 169(6):1439–1441, 1997. doi: 10.2214/ajr.169.6.9393162. URL https: //www.ajronline.org/doi/10.2214/ajr.169.6.9393162

    work page doi:10.2214/ajr.169.6.9393162 1997

  7. [8]

    R. A. Lerski and J. D. de Certaines. Performance assessment and quality control in mri by eurospin test objects and protocols.Magnetic Resonance Imaging, 11(6):817–833, 1993. doi: 10.1016/0730-725X(93)90199-N. URLhttps://pubmed.ncbi.nlm.nih.gov/8371637/

    work page doi:10.1016/0730-725x(93)90199-n 1993

  8. [9]

    Deep learning: A primer for radiologists.RadioGraphics, 43(1):e1–e15, 2023

    Shingo Kiryu, Hiroyuki Abe, Yusuke Hara, Takayuki Shimizu, Yoshitaka Narita, and Hiroshi Fujita. Deep learning: A primer for radiologists.RadioGraphics, 43(1):e1–e15, 2023. doi: 10.1148/rg.211719. URLhttps://pubs.rsna.org/doi/10.1148/rg.211719

    work page doi:10.1148/rg.211719 2023

  9. [11]

    Toews, Philip K

    Alexander R. Toews, Philip K. Lee, Krishna S. Nayak, and Brian A. Hargreaves. Comprehen- sive assessment of nonuniform image quality: Application to imaging near metal.Magnetic Resonance in Medicine, 92(6):2358–2372, 2024. doi: https://doi.org/10.1002/mrm.30222. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/mrm.30222

    work page doi:10.1002/mrm.30222 2024

  10. [13]

    Sodickson, Mark A

    Daniel K. Sodickson, Mark A. Griswold, Peter M. Jakob, Robert R. Edelman, and Warren J. Manning. Signal-to-noise ratio and signal-to-noise efficiency in smash imaging.Magnetic Reso- nance in Medicine, 41:1009–1022, 1999. ISSN 07403194. doi: 10.1002/(SICI)1522-2594(199905) 41:5<1009::AID-MRM21>3.0.CO;2-4

    work page doi:10.1002/(sici)1522-2594(199905 1999

  11. [14]

    Breuer, Stephan A.R

    Felix A. Breuer, Stephan A.R. Kannengiesser, Martin Blaimer, Nicole Seiberlich, Peter M. Jakob, and Mark A. Griswold. General formulation for quantitative g-factor calculation in GRAPPA reconstructions.Magnetic Resonance in Medicine, 62(3):739–746, 2009. doi: 10.1002/mrm.22066

    work page doi:10.1002/mrm.22066 2009

  12. [15]

    Yakushiji, Ichiro Tani, Yasuo Nakajima, and Marc Van Cauteren

    Yasuyuki Kurihara, Yoshiko K. Yakushiji, Ichiro Tani, Yasuo Nakajima, and Marc Van Cauteren. Coil sensitivity encoding in mr imaging.American Journal of Roentgenology, 178(5):1087–1091,

    work page

  13. [16]

    URL https://doi.org/10.2214/ajr.178.5.1781087

    doi: 10.2214/ajr.178.5.1781087. URL https://doi.org/10.2214/ajr.178.5.1781087. PMID: 11959706

    work page doi:10.2214/ajr.178.5.1781087

  14. [17]

    Goerner and Geoffrey D

    Frank L. Goerner and Geoffrey D. Clarke. Measuring signal-to-noise ratio in partially parallel imaging mri.Medical Physics, 38:5049–5057, 2011. ISSN 00942405. doi: 10.1118/1.3618730

    work page doi:10.1118/1.3618730 2011

  15. [18]

    Pruessmann, Markus Weiger, Peter Börnert, and Peter Boesiger

    Klaas P. Pruessmann, Markus Weiger, Peter Börnert, and Peter Boesiger. Advances in sensitivity encoding with arbitrary k-space trajectories.Magnetic Resonance in Medicine, 46:638–651,

    work page

  16. [19]

    doi: 10.1002/mrm.1241

    ISSN 07403194. doi: 10.1002/mrm.1241

    work page doi:10.1002/mrm.1241

  17. [20]

    Augmented generalized sense reconstruction to correct for rigid body motion.Magnetic Resonance in Medicine, 57(1):90–102, 2007

    Roland Bammer, Murat Aksoy, and Chunlei Liu. Augmented generalized sense reconstruction to correct for rigid body motion.Magnetic Resonance in Medicine, 57(1):90–102, 2007. doi: 19 https://doi.org/10.1002/mrm.21106. URL https://onlinelibrary.wiley.com/doi/abs/10. 1002/mrm.21106

    work page doi:10.1002/mrm.21106 2007

  18. [21]

    Cg-sense revisited: Results from the first ismrm reproducibility challenge.Magnetic Resonance in Medicine, 85(4):1821–1839, 2021

    Oliver Maier, Steven Hubert Baete, Alexander Fyrdahl, Kerstin Hammernik, Seb Harrevelt, Lars Kasper, Agah Karakuzu, Michael Loecher, Franz Patzig, Ye Tian, Ke Wang, Daniel Gallichan, Martin Uecker, and Florian Knoll. Cg-sense revisited: Results from the first ismrm reproducibility challenge.Magnetic Resonance in Medicine, 85(4):1821–1839, 2021. doi: https...

    work page doi:10.1002/mrm.28569 2021

  19. [22]

    Compressed sensing mri: a review of the clinical literature.British Journal of Radiology, 88(1056):20150487, 10 2015

    Oren N Jaspan, Roman Fleysher, and Michael L Lipton. Compressed sensing mri: a review of the clinical literature.British Journal of Radiology, 88(1056):20150487, 10 2015. ISSN 0007-1285. doi: 10.1259/bjr.20150487. URLhttps://doi.org/10.1259/bjr.20150487

    work page doi:10.1259/bjr.20150487 2015

  20. [23]

    Vasanawala, Marcus T

    Shreyas S. Vasanawala, Marcus T. Alley, Brian A. Hargreaves, Richard A. Barth, John M. Pauly, and Michael Lustig. Improved pediatric mr imaging with compressed sensing.Radiology, 256(2):607–616, 2010. doi: 10.1148/radiol.10091218. URLhttps://doi.org/10.1148/radiol. 10091218. PMID: 20529991

    work page doi:10.1148/radiol.10091218 2010

  21. [24]

    Physics-driven deep learning for computational magnetic resonance imaging: Combining physics and machine learning for improved medical imaging

    Kerstin Hammernik, Thomas Küstner, Burhaneddin Yaman, Zhengnan Huang, Daniel Rueckert, Florian Knoll, and Mehmet Akçakaya. Physics-driven deep learning for computational magnetic resonance imaging: Combining physics and machine learning for improved medical imaging. IEEE Signal Processing Magazine, 40(1):98–114, 2023. doi: 10.1109/MSP.2022.3215288

    work page doi:10.1109/msp.2022.3215288 2023

  22. [25]

    Deep learning for accelerated and robust MRI reconstruction.Magnetic Resonance Materials in Physics, Biology and Medicine, 37(3):335–368, July 2024

    Reinhard Heckel, Mathews Jacob, Akshay Chaudhari, Or Perlman, and Efrat Shimron. Deep learning for accelerated and robust MRI reconstruction.Magnetic Resonance Materials in Physics, Biology and Medicine, 37(3):335–368, July 2024. doi: 10.1007/s10334-024-01173-8. URLhttps://doi.org/10.1007/s10334-024-01173-8

    work page doi:10.1007/s10334-024-01173-8 2024

  23. [27]

    Fessler, Sangwoo Lee, V.T

    J.A. Fessler, Sangwoo Lee, V.T. Olafsson, H.R. Shi, and D.C. Noll. Toeplitz-based iterative im- age reconstruction for mri with correction for magnetic field inhomogeneity.IEEE Transactions on Signal Processing, 53(9):3393–3402, 2005. doi: 10.1109/TSP.2005.853152

    work page doi:10.1109/tsp.2005.853152 2005

  24. [28]

    Noise distribution in sense- and grappa-reconstructed images: a computer simulation study.Magnetic Resonance Imaging, 25:1089–1094, 9 2007

    Per Thunberg and Per Zetterberg. Noise distribution in sense- and grappa-reconstructed images: a computer simulation study.Magnetic Resonance Imaging, 25:1089–1094, 9 2007. ISSN 0730725X. doi: 10.1016/j.mri.2006.11.003

    work page doi:10.1016/j.mri.2006.11.003 2007

  25. [29]

    Seeking a widely adoptable practical standard to estimate signal-to-noise ratio in magnetic resonance imaging for multiple-coil reconstructions.J

    Eros Montin and Riccardo Lattanzi. Seeking a widely adoptable practical standard to estimate signal-to-noise ratio in magnetic resonance imaging for multiple-coil reconstructions.J. Magn. Reson. Imaging, 54(6):1952–1964, 2021. doi: 10.1002/jmri.27816

    work page doi:10.1002/jmri.27816 1952

  26. [31]

    Noise-induced variability quantification in deep learning-based mri reconstructions

    Onat Dalmaz, Arjun Divyang Desai, Akshay Chaudhari, and Brian Hargreaves. Noise-induced variability quantification in deep learning-based mri reconstructions. In2024 ISMRM and ISMRT Annual Meeting and Exhibition, Singapore, Singapore, May 2024. ISMRM. URLhttp: //echo.ismrm.org/abstracts/view/4d1e145f-9b3a-4949-9121-1f5d15f99c48. Program Number: 2785

    work page 2024

  27. [32]

    Basha, Warren J

    Mehmet Akcakaya, Tamer A. Basha, Warren J. Manning, and Reza Nezafat. Efficient calculation of g-factors for cg-sense in high dimensions: noise amplification in random undersampling. Journal of Cardiovascular Magnetic Resonance, 16, 1 2014. ISSN 1532429X. doi: 10.1186/ 1532-429X-16-S1-W28

    work page 2014

  28. [33]

    Kerr, Petter Dyverfeldt, Charles J

    Pierre Daudé, Susan Schnell, Caroline Wu, Aurelien B. Kerr, Petter Dyverfeldt, Charles J. François, Michael Markl, Brian A. Hargreaves, and Krishna S. Nayak. Inline automatic quality control of 2D phase-contrast flow MR imaging for subject-specific scan time adaptation.Magn. Reson. Med., 92(2):908–922, 2024. doi: 10.1002/mrm.30083

    work page doi:10.1002/mrm.30083 2024

  29. [34]

    Wiens, Shawn J

    Curtis N. Wiens, Shawn J. Kisch, Jacob D. Willig-Onwuachi, and Charles A. McKenzie. Computationally rapid method of estimating signal-to-noise ratio for phased array image reconstructions.Magn. Reson. Med., 66(4):1192–1197, 2011. doi: 10.1002/mrm.22893

    work page doi:10.1002/mrm.22893 2011

  30. [35]

    Efficient noise calculation in deep learning-based MRI reconstructions

    Onat Dalmaz, Arjun D Desai, Reinhard Heckel, Tolga Cukur, Akshay S Chaudhari, and Brian Hargreaves. Efficient noise calculation in deep learning-based MRI reconstructions. In Aarti Singh, Maryam Fazel, Daniel Hsu, Simon Lacoste-Julien, Felix Berkenkamp, Tegan Maharaj, Kiri Wagstaff, and Jerry Zhu, editors,Proceedings of the 42nd International Conference o...

    work page 2025

  31. [36]

    Dawood, F

    P. Dawood, F. Breuer, M. Gram, I. Homolya, P. M. Jakob, M. Zaiss, and M. Blaimer. Image space formalism of convolutional neural networks for k-space interpolation.Magnetic Resonance in Medicine, 94(6):2680–2701, 2025. doi: https://doi.org/10.1002/mrm.70002. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/mrm.70002

    work page doi:10.1002/mrm.70002 2025

  32. [37]

    Controlling spatial correlation in k-space interpolation networks for MRI reconstruction: denoising versus apparent blurring.arXiv preprint arXiv:2505.11155, 2025

    Istvan Homolya, Jannik Stebani, Felix Breuer, Grit Hein, Matthias Gamer, Florian Knoll, and Martin Blaimer. Controlling spatial correlation in k-space interpolation networks for MRI reconstruction: denoising versus apparent blurring.arXiv preprint arXiv:2505.11155, 2025. doi: 10.48550/arXiv.2505.11155

    work page doi:10.48550/arxiv.2505.11155 2025

  33. [38]

    Hutchinson

    Michael F. Hutchinson. A stochastic estimator of the trace of the influence matrix for laplacian smoothing splines.J. Commun. Statist. Simula., 19(2):433–450, 1990. URLhttps://cir.nii. ac.jp/crid/1573668925763299072

    work page arXiv 1990

  34. [39]

    Bekas, E

    C. Bekas, E. Kokiopoulou, and Y. Saad. An estimator for the diagonal of a matrix.Applied Numerical Mathematics, 57(11):1214–1229, 2007. ISSN 0168-9274. doi: https://doi.org/10. 1016/j.apnum.2007.01.003. URL https://www.sciencedirect.com/science/article/pii/ S0168927407000244. Numerical Algorithms, Parallelism and Applications (2)

    work page 2007

  35. [40]

    Hansen and Peter Kellman

    Michael S. Hansen and Peter Kellman. Image reconstruction: an overview for clinicians.J. Magn. Reson. Imaging, 41(3):573–585, 2015. doi: 10.1002/jmri.24687

    work page doi:10.1002/jmri.24687 2015

  36. [41]

    Algebraic 21 methods and computational strategies for pseudoinverse-based mr image reconstruction (pinv- recon).Scientific Reports, 15(1):37997, 2025

    Kylie Yeung, Christine Tobler, Rolf F Schulte, Benjamin White, Anthony McIntyre, Sébastien Serres, Peter Morris, Dorothee Auer, Fergus V Gleeson, Damian J Tyler, et al. Algebraic 21 methods and computational strategies for pseudoinverse-based mr image reconstruction (pinv- recon).Scientific Reports, 15(1):37997, 2025

    work page 2025

  37. [42]

    Donoho, Juan M

    Michael Lustig, David L. Donoho, Juan M. Santos, and John M. Pauly. Compressed sensing mri.IEEE Signal Processing Magazine, 25(2):72–82, 2008. doi: 10.1109/MSP.2007.914728

    work page doi:10.1109/msp.2007.914728 2008

  38. [43]

    Creation of fully sampled mr data repository for compressed sensing of the knee

    Kevin Epperson, Anne Marie Sawyer, Michael Lustig, Marcus Alley, Martin Uecker, Patrick Virtue, Peng Lai, and Shreyas Vasanawala. Creation of fully sampled mr data repository for compressed sensing of the knee. ISMRM 2013 Meeting Proceedings. Section for Magnetic Resonance Technologists, 2013

    work page 2013

  39. [44]

    Joint image reconstruction and sensitivity estimation in sense (jsense).Magnetic Resonance in Medicine, 57:1196–1202, 2007

    Leslie Ying and Jinhua Sheng. Joint image reconstruction and sensitivity estimation in sense (jsense).Magnetic Resonance in Medicine, 57:1196–1202, 2007. ISSN 15222594. doi: 10.1002/mrm.21245

    work page doi:10.1002/mrm.21245 2007

  40. [45]

    Murphy, Patrick Virtue, Michael Elad, John M

    Martin Uecker, Peng Lai, Mark J. Murphy, Patrick Virtue, Michael Elad, John M. Pauly, Shreyas S. Vasanawala, and Michael Lustig. Espirit - an eigenvalue approach to autocalibrating parallel mri: Where sense meets grappa.Magnetic Resonance in Medicine, 71:990–1001, 2014. ISSN 15222594. doi: 10.1002/mrm.24751

    work page doi:10.1002/mrm.24751 2014

  41. [46]

    Muckley, Mary Bruno, Aaron Defazio, Marc Parente, Krzysztof J

    Florian Knoll, Jure Zbontar, Anuroop Sriram, Matthew J. Muckley, Mary Bruno, Aaron Defazio, Marc Parente, Krzysztof J. Geras, Joe Katsnelson, Hersh Chandarana, Zizhao Zhang, Michal Drozdzalv, Adriana Romero, Michael Rabbat, Pascal Vincent, James Pinkerton, Duo Wang, Nafissa Yakubova, Erich Owens, C. Lawrence Zitnick, Michael P. Recht, Daniel K. Sodickson,...

    work page doi:10.1148/ryai.2020190007 2020

  42. [47]

    A fast iterative shrinkage-thresholding algorithm for linear inverse problems

    AmirBeckandMarcTeboulle. Afastiterativeshrinkage-thresholdingalgorithmforlinearinverse problems.SIAM Journal on Imaging Sciences, 2(1):183–202, 2009. doi: 10.1137/080716542

    work page doi:10.1137/080716542 2009

  43. [48]

    Second order total generalized variation (tgv) for mri.Magnetic Resonance in Medicine, 65(2):480–491, 2011

    Florian Knoll, Kristian Bredies, Thomas Pock, and Rudolf Stollberger. Second order total generalized variation (tgv) for mri.Magnetic Resonance in Medicine, 65(2):480–491, 2011. doi: https://doi.org/10.1002/mrm.22595. URL https://onlinelibrary.wiley.com/doi/abs/10. 1002/mrm.22595

    work page doi:10.1002/mrm.22595 2011

  44. [49]

    Curran Associates Inc., Red Hook, NY, USA, 2019

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Köpf, Edward Yang, Zach DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu Fang, Junjie Bai, and Soumith Chintala.PyTorch: an imperative style, high-performan...

    work page 2019

  45. [50]

    Sigpy: A python package for high performance iterative reconstruction.Proc

    Frank Ong and Michael Lustig. Sigpy: A python package for high performance iterative reconstruction.Proc. Intl. Soc. Mag. Reson. Med., 27:4819, 2019

    work page 2019

  46. [51]

    Brown, Y.C.N

    R.W. Brown, Y.C.N. Cheng, E.M. Haacke, M.R. Thompson, and R. Venkatesan.Magnetic Resonance Imaging: Physical Principles and Sequence Design. Wiley, 2014. ISBN9781118633977. URLhttps://books.google.com/books?id=rQGCAwAAQBAJ

    work page 2014

  47. [52]

    Wiley-Blackwell, 5th edition, 2018

    Catherine Westbrook, Carolyn Kaut Roth, and John Talbot.MRI in Practice. Wiley-Blackwell, 5th edition, 2018. ISBN 978-1119391968. 22

    work page 2018

  48. [53]

    Randomized algorithms for estimating the trace of an implicit symmetric positive semi-definite matrix.Journal of the ACM, 58(2):8, 2011

    Haim Avron and Sivan Toledo. Randomized algorithms for estimating the trace of an implicit symmetric positive semi-definite matrix.Journal of the ACM, 58(2):8, 2011. doi: 10.1145/ 1944345.1944349

    work page arXiv 2011

  49. [54]

    Robson, Aaron K

    Philip M. Robson, Aaron K. Grant, Ananth J. Madhuranthakam, Riccardo Lattanzi, Daniel K. Sodickson, and Charles A. McKenzie. Comprehensive quantification of signal-to-noise ratio and g-factor for image-based and k-space-based parallel imaging reconstructions.Magnetic Resonance in Medicine, 60(4):895–907, 2008. doi: 10.1002/mrm.21728

    work page doi:10.1002/mrm.21728 2008

  50. [55]

    Sodickson, and Mehmet Akcakaya

    Florian Knoll, Kerstin Hammernik, Chi Zhang, Steen Moeller, Thomas Pock, Daniel K. Sodickson, and Mehmet Akcakaya. Deep-learning methods for parallel magnetic resonance imaging reconstruction: a survey of the current approaches, trends, and issues.IEEE Signal Process. Mag., 37(1):128–140, 2020. doi: 10.1109/MSP.2019.2950640

    work page doi:10.1109/msp.2019.2950640 2020

  51. [56]

    Data-driven regularization parameter selection in dynamic MRI.J

    Matti Hanhela, Mikael Delić, Aku Nissinen, and Ville Kolehmainen. Data-driven regularization parameter selection in dynamic MRI.J. Imaging, 7(2):38, 2021. doi: 10.3390/jimaging7020038

    work page doi:10.3390/jimaging7020038 2021

  52. [57]

    An analytical solution to the dispersion-by-inversion problem in magnetic resonance elastography.Magnetic Res- onance in Medicine, 84(1):61–71, July 2020

    Mathieu Boudreau, Agah Karakuzu, Julien Cohen-Adad, Ecem Bozkurt, Madeline Carr, Marco Castellaro, Luis Concha, Mariya Doneva, Seraina A. Dual, Alex Ensworth, Alexandru Foias, Véronique Fortier, Refaat E. Gabr, Guillaume Gilbert, Carri K. Glide-Hurst, Matthew Grech-Sollars, Siyuan Hu, Oscar Jalnefjord, Jorge Jovicich, Kübra Keskin, Peter Koken, Anastasia ...

    work page doi:10.1002/mrm 2024

  53. [58]

    Siria Pasini, Steffen Ringgaard, Tau Vendelboe, Leyre Garcia-Ruiz, Anika Strittmatter, Giulia Villa, Anish Raj, Rebeca Echeverria-Chasco, Michela Bozzetto, Paolo Brambilla, Malene Aastrup, Esben S. S. Hansen, Luisa Pierotti, Matteo Renzulli, Susan T. Francis, Frank G. Zöllner, Christoffer Laustsen, Maria A. Fernandez-Seara, and Anna Caroli. Multi-center a...

    work page doi:10.1007/s10334-025-01260-4 2025

  54. [59]

    MR Optimum: A web-based open- source tool for standardized signal-to-noise ratio evaluation in MRI.Computer Methods and Programs in Biomedicine Update, page 100235, 2026

    Eros Montin, Xuan Thao Nguyen, and Riccardo Lattanzi. MR Optimum: A web-based open- source tool for standardized signal-to-noise ratio evaluation in MRI.Computer Methods and Programs in Biomedicine Update, page 100235, 2026. doi: 10.1016/j.cmpbup.2026.100235

    work page doi:10.1016/j.cmpbup.2026.100235 2026

  55. [60]

    Hess, Catherine J

    Jeremiah J. Hess, Catherine J. Moran, Preya Shah, Jana Vincent, Fraser J. L. Robb, Bruce L. Daniel, and Brian A. Hargreaves. Relative SNR measurements in supine vs. prone breast MRI. Magnetic Resonance in Medicine, 95(5):2718–2725, 2026. doi: 10.1002/mrm.70217

    work page doi:10.1002/mrm.70217 2026

  56. [61]

    Detection of brain functional-connectivity difference in post-stroke patients using group-level covariance modeling

    Gaël Varoquaux, Flore Baronnet, Andreas Kleinschmidt, Pierre Fillard, and Bertrand Thirion. Detection of brain functional-connectivity difference in post-stroke patients using group-level covariance modeling. In Tianzi Jiang, Nassir Navab, Josien P. W. Pluim, and Max A. Viergever, 23 editors,Medical Image Computing and Computer-Assisted Intervention – MIC...

    work page 2010

  57. [62]

    Unsupervised representation learning of brain activity via bridging voxel activity and functional connectivity

    Ali Behrouz, Parsa Delavari, and Farnoosh Hashemi. Unsupervised representation learning of brain activity via bridging voxel activity and functional connectivity. InProceedings of the 41st International Conference on Machine Learning (ICML), volume 235 ofProceedings of Machine Learning Research, pages 3347–3381, 2024

    work page 2024

  58. [63]

    Methods for uncertainty quantification in dictionary matching to advance interpretable quantitative MRI

    Brian Toner. Methods for uncertainty quantification in dictionary matching to advance interpretable quantitative MRI. Sedona, AZ, United States of America, January 2026. Proc. Intl. Soc. Mag. Reson. Med. Workshop on Data Sampling and Image Reconstruction. URL https://echo.ismrm.org/program/SEDONA26. Abstract #00187

    work page 2026

  59. [64]

    Hanchate and Kalyani R

    Vandana V. Hanchate and Kalyani R. Joshi. Mri denoising using bm3d equipped with noise invalidation denoising technique and vst for improved contrast.SN Applied Sciences, 2(2):234, 2020

    work page 2020

  60. [65]

    Trzasko, David S

    Zhoubo Li, Lifeng Yu, Joshua D. Trzasko, David S. Lake, Daniel J. Blezek, Joel G. Fletcher, Cynthia H. McCollough, and Armando Manduca. Adaptive nonlocal means filtering based on local noise level for ct denoising.Medical Physics, 41(1):011908, 2014. doi: 10.1118/1.4851635

    work page doi:10.1118/1.4851635 2014

  61. [66]

    Data-driven estimation of noise variance stabilization parameters for low-dose x-ray images.Physics in Medicine and Biology, 65(22):225027, 2020

    Sai Gokul Hariharan, Norbert Strobel, Christian Kaethner, Markus Kowarschik, Rebecca Fahrig, and Nassir Navab. Data-driven estimation of noise variance stabilization parameters for low-dose x-ray images.Physics in Medicine and Biology, 65(22):225027, 2020. doi: 10.1088/ 1361-6560/abbc82

    work page 2020

  62. [67]

    Yongjian Yu and Scott T. Acton. Speckle reducing anisotropic diffusion.IEEE Transactions on Image Processing, 11(11):1260–1270, 2002. doi: 10.1109/TIP.2002.804276

    work page doi:10.1109/tip.2002.804276 2002

  63. [68]

    Campbell-Washburn, John P

    Quan Dou, Zhixing Wang, Xue Feng, Adrienne E. Campbell-Washburn, John P. Mugler, and Craig H. Meyer. Mri denoising with a non-blind deep complex-valued convolutional neural network.NMR in Biomedicine, 38(1):e5291, 2025. doi: 10.1002/nbm.5291

    work page doi:10.1002/nbm.5291 2025

  64. [69]

    U-medsam: Uncertainty-aware medsam for medical image segmentation

    Xin Wang, Xiaoyu Liu, Peng Huang, Pu Huang, Shu Hu, and Hongtu Zhu. U-medsam: Uncertainty-aware medsam for medical image segmentation. In Jun Ma, Yuyin Zhou, and Bo Wang, editors,Medical Image Segmentation Foundation Models. CVPR 2024 Challenge: Segment Anything in Medical Images on Laptop, pages 206–217, Cham, 2025. Springer Nature Switzerland. ISBN 978-...

    work page 2024

  65. [70]

    SNRAware: Improved Deep Learning MRI Denoising with Signal-to-Noise Ratio Unit Training and G-Factor Map Augmentation,

    Hui Xue, Sarah M. Hooper, Iain Pierce, Rhodri H. Davies, John Stairs, Joseph Naegele, Adrienne E. Campbell-Washburn, Charlotte Manisty, James C. Moon, Thomas A. Treibel, Michael S. Hansen, and Peter Kellman. SNRAware: Improved deep learning MRI denoising with signal-to-noise ratio unit training and g-factor map augmentation.Radiology: Artificial Intellige...

    work page doi:10.1148/ryai.250227 2025

  66. [71]

    Pa- tricia Bandettini, Rajiv Ramasawmy, Ahsan Javed, Zheren Zhu, Yang Yang, James Moon, Adrienne Campbell, and Peter Kellman

    Hui Xue, Sarah Hooper, Azaan Rehman, Iain Pierce, Thomas Treibel, Rhodri Davies, W. Pa- tricia Bandettini, Rajiv Ramasawmy, Ahsan Javed, Zheren Zhu, Yang Yang, James Moon, Adrienne Campbell, and Peter Kellman. Imaging transformer for MRI denoising with the SNR unit training: enabling generalization across field-strengths, imaging contrasts, and anatomy,

    work page

  67. [72]

    arXiv:2404.02382. 24

    work page arXiv

  68. [73]

    Pruessmann, Markus Weiger, Markus B

    Klaas P. Pruessmann, Markus Weiger, Markus B. Scheidegger, and Peter Boesiger. Sense: Sensitivity encoding for fast mri.Magnetic Resonance in Medicine, 42(5):952–962, 1999. doi: 10.1002/(SICI)1522-2594(199911)42:5<952::AID-MRM16>3.0.CO;2-S

    work page doi:10.1002/(sici)1522-2594(199911)42:5 1999

  69. [74]

    Harrison H. Barrett. Task-based measures of image quality and their relation to radiation dose and patient risk.Physics in Medicine and Biology, 60(2):R1–R75, 2015. doi: 10.1088/ 0031-9155/60/2/R1

    work page 2015

  70. [75]

    Sergey Kastryulin, Jamil Zakirov, Nicola Pezzotti, and Dmitry V. Dylov. Image quality assessment for magnetic resonance imaging.IEEE Access, 11:14154–14168, 2023. doi: 10.1109/ ACCESS.2023.3243466

    work page arXiv 2023

  71. [76]

    Rad-IQMRI: A benchmark for MRI image quality assessment.Neurocomputing, 602:128292, 2024

    Yueran Ma, Jianxun Lou, Jean-Yves Tanguy, Padraig Corcoran, and Hantao Liu. Rad-IQMRI: A benchmark for MRI image quality assessment.Neurocomputing, 602:128292, 2024. doi: 10.1016/j.neucom.2024.128292

    work page doi:10.1016/j.neucom.2024.128292 2024

  72. [77]

    Eisenmenger, Leonardo Rivera-Rivera, Eugene Huo, Jacqueline C

    Chenwei Tang, Laura B. Eisenmenger, Leonardo Rivera-Rivera, Eugene Huo, Jacqueline C. Junn, Anthony D. Kuner, Thekla H. Oechtering, Anthony Peret, Jitka Starekova, and Kevin M. Johnson. Incorporating radiologist knowledge into MRI quality metrics for machine learning using rank-based ratings.Journal of Magnetic Resonance Imaging, 61(6):2572–2584, 2025. do...

    work page doi:10.1002/jmri.29672 2025

  73. [78]

    Cambridge Series in Statistical and Probabilistic Mathematics

    Roman Vershynin.High-Dimensional Probability: An Introduction with Applications in Data Science. Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press, Cambridge, 2 edition, 2026. 25 Algorithm 1Probing Image-space COvariance (PICO) Require: Encoding operatorA(and adjointA H); regularization parameterλ≥0; number of probe...

    work page 2026