Pith. sign in

REVIEW 4 major objections 5 minor 44 references

Emulating Clinical Quality Muscle B-mode Ultrasound Images from Plane Wave Images Using a Two-Stage Machine Learning Model

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A two-stage neural network turns single plane-wave frames into clinical-style muscle images.

desk verdict A practical, well-scoped two-stage ML pipeline for real-time muscle plane-wave-to-clinical-style ultrasound, with code and a statistically significant reader study; the 'clinical quality' framing is a bit stronger than the evidence, but the paper deserves refereeing. read the letter →

arxiv 2412.05758 v1 pith:XPQGLKQZ submitted 2024-12-07 eess.IV cs.CVcs.LG

Reed Chen , Courtney Trutna Paley , Wren Wightman , Lisa Hobson-Webb , Yohei Harada , Felix Jin , Ouwen Huang , Mark Palmeri
show 1 more author
Kathryn Nightingale
This is my paper · ORCID
classification eess.IVcs.CVcs.LG
keywords planewaveimagingmuscleultrasoundCycleGANU-Netshearelasticityreal-timereaderstudyimageenhancement
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper aims to show that a low-quality single plane-wave ultrasound frame of skeletal muscle, the kind used in shear wave elasticity imaging, can be upgraded in real time into an image that looks like a clinical scanner's output. The proposed route has two stages: a U-Net learns to mimic the effect of compounding several plane waves and applying standard post-processing, and a CycleGAN then learns the visual style of clinical muscle B-modes from unpaired example images. The authors support the claim with image metrics and a reader study in which two physicians scored the final images as having significantly less speckle and higher structural fidelity than the raw plane wave frames. If the claim holds, researchers can watch a continuously updated, clinically interpretable B-mode during shear wave acquisitions without interrupting the high-frame-rate sequence.

What carries the argument

The load-bearing mechanism is the two-stage cascade. Stage one is a U-Net, sharing the CycleGAN's generator architecture, trained with L1 loss on paired images to map a single 0-degree plane wave image to the output of 12-angle coherent plane wave compounding plus histogram matching and unsharp masking. Stage two is a CycleGAN with two generators and two PatchGAN discriminators, trained with cycle-consistency, identity, and least-squares adversarial losses, that translates stage-one outputs into the clinical muscle domain. The real-time result comes from exporting the trained networks and running them inside the scanner's acquisition loop, so each single plane wave transmit produces a continuously refreshed clinical-style display.

What would settle it

Run the current two-stage model on a test set of longitudinal muscle images only and have a blinded panel of physicians score structural fidelity against matched clinical longitudinal B-modes. If the longitudinal-only scores are significantly worse than the pooled reader-study scores, or if the outputs show transverse starry-night texture on longitudinal anatomy, then the mixed orientation-imbalanced clinical domain does not define a single emulatable clinical style.

Watch

Extended reading notes

Core claim

A single unfocused plane wave transmit can be mapped, in two learned steps, to an image that physician readers judge to be of clinical quality. Stage one reproduces the target formed by 12-angle coherent plane wave compounding followed by histogram matching and unsharp masking; stage two transfers the result into the appearance of clinical muscle B-modes. On test ROIs, the cascade lowers speckle standard deviation from 0.208 to 0.068, raises contrast-to-noise ratio from 3.67 to 5.11, and lowers fiber/fascicle line standard deviation from 0.162 to 0.115. In the reader study, the stage-2 output differed significantly from the plane wave input on both speckle and structural fidelity, while the stage-1 output was statistically indistinguishable from the compounded-and-filtered reference, supporting the claim that stage one learned its target. The whole stack runs at 28.5 ± 0.6 frames per second on the scanner.

Load-bearing premise

The model assumes that clinical muscle images collected from several different scanners, mostly showing transverse views, form one consistent clinical style that the network can learn to copy.

Editorial extensions

If this is right

  • Single-plane-wave shear wave elasticity acquisitions can show a real-time B-mode-like view without adding separate B-mode sequences, simplifying probe positioning during the acquisition.
  • The two-stage outputs on test images improved speckle, contrast-to-noise ratio, and fiber cohesiveness in the measured ROIs.
  • The first stage alone reproduces the compounded-and-filtered reference at about 18 ms per image on a GPU, roughly an order of magnitude faster than the traditional CPU pipeline it emulates.
  • The two-stage stack achieves 28.5 ± 0.6 FPS on the scanner, which the authors classify as real time.
  • Because the stage-2 network can connect or create structures, the two-stage design also gives operators a less processed first-stage view that stays closer to the original data.

Reading between the lines

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

  • The same cascade could transfer to other shear wave targets, such as liver or thyroid, because nothing in the method is muscle-specific except the training data.
  • The clinical training set's imbalance (2,354 transverse versus 1,146 longitudinal images) suggests a sharper test than the pooled reader study: evaluating stage-2 outputs on longitudinal-only test images may reveal whether the model imposes a transverse 'starry night' texture on longitudinal anatomy.
  • The asymmetric generator losses chosen for the CycleGAN mean the reverse translation is not held to the same standard, so the model's behavior on the clinical-to-research direction is largely untested and could differ substantially.
  • Stage one's speed and fidelity to the compounding-plus-filtering pipeline make it a candidate low-latency replacement for that pipeline in other research-scanner workflows, independent of the clinical-style stage.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. This paper presents a two-stage machine learning pipeline for enhancing single-plane-wave ultrasound images of skeletal muscle acquired on a Verasonics research scanner. The first stage is a U-Net trained with paired data to emulate plane-wave compounding, histogram matching, and unsharp masking. The second stage is a CycleGAN trained with unpaired data to translate the stage-1 output into images that resemble clinical muscle B-modes, using a heterogeneous collection of four online repositories. The pipeline is implemented on a Verasonics system and achieves 28.5 ± 0.6 FPS. Evaluation includes quantitative image metrics (speckle standard deviation, CNR, and fiber/fascicle standard deviation) and a reader study with two neurologists. The reader study reports that the stage-2 output has significantly lower speckle and higher structural fidelity than the original plane-wave image, the PWC+filtered image, and the stage-1 output, based on Friedman's ANOVA and Nemenyi post-hoc tests. The authors acknowledge limitations including CycleGAN hallucination, an imbalanced transverse/longitudinal training set, and a heterogeneous target domain.

Significance. If the claims are substantiated, the paper offers a practical real-time enhancement pathway for plane-wave muscle ultrasound, potentially useful in shear-wave elastography workflows where continuous B-mode feedback is valuable. The two-stage design and the detailed Verasonics/TensorFlow implementation are useful engineering contributions, and the code is made available. The quantitative and reader-based evaluations are a reasonable first step. However, the central claim of emulating 'clinical quality' is not directly validated: no actual clinical B-mode image is used as a reference in the reader study or in the quantitative evaluation, and the 'structural fidelity' metric is a subjective clarity rating rather than a measure of anatomical correctness. The small reader pool and the heterogeneous, orientation-imbalanced target domain further limit the strength of the conclusions. The paper is a useful feasibility demonstration, but the headline claim requires additional evaluation or a more cautious framing.

major comments (4)
  1. [Section II-E, Section III-C, Section IV] The reader study evaluates only the four research-scanner variants (plane-wave input, PWC+filter, stage 1, stage 2) and does not include any actual clinical B-mode image. The abstract's claim that the processed images have 'significantly greater structural fidelity' is therefore a claim about perceived clarity relative to the research images, not a demonstration that the outputs match clinical muscle B-modes. This distinction matters because Section IV states that the stage-2 CycleGAN 'can also generate structures that were not originally present'; higher reader-rated structural fidelity might reflect plausible-looking hallucinated fibers rather than true preservation of anatomy. The authors should either include clinical B-mode images (ideally of the same or similar anatomy) in the reader study, or revise the central claim to 'perceived structural clarity' to avoid overstatement.
  2. [Section II-E, Section III-C, Figure 9] The reader study relies on only two readers, and their scores are averaged before the Friedman ANOVA and Nemenyi post-hoc tests are computed, making the 20 image sets the statistical unit and ignoring between-reader variability. The power analysis is described only by an assumed Cohen's effect size of 0.35; the target power, alpha, and the number of readers are not specified. With two readers from the same institution, the significant p-values do not provide strong evidence of generalizable reader preference. The authors should report per-reader scores, quantify inter-reader agreement (e.g., weighted kappa or ICC), and ideally increase the number of readers, or at minimum temper the generality of the reader-study conclusion.
  3. [Section II-B, Table I, Section IV] The CycleGAN's target 'clinical muscle B-mode' domain is a mixture of four online repositories with different scanners, resolutions, and dynamic ranges, and is heavily dominated by transverse images (2,354 transverse vs 1,146 longitudinal). As the authors acknowledge in Section IV, this imbalance may bias the model toward transverse-like outputs even for longitudinal inputs. Since the central claim is that the model emulates clinical muscle B-modes, the target style is not well defined. The authors should either train on a more homogeneous clinical dataset or evaluate transverse and longitudinal outputs separately, ideally comparing the translations to actual clinical images of the corresponding orientation.
  4. [Table III, Section IV] The quantitative metrics in Table III are not independent of the hallucination mechanism. In particular, the fiber/fascicle standard deviation measures the continuity of hyperechoic lines; the CycleGAN reduces this metric by connecting interrupted fibers and fascicles, which is precisely the behavior that Section IV acknowledges can 'generate structures that were not originally present.' Similarly, speckle standard deviation and CNR measure textural properties, not anatomical correctness. The reported improvements in these metrics therefore do not establish that the processed images are more clinically faithful; they only show that the images appear smoother and have higher contrast. The authors should either validate against independent anatomical ground truth or explicitly restrict their conclusions to appearance rather than clinical fidelity.
minor comments (5)
  1. [Section I] The sentence 'This makes fDAS-US less unsuitable for applications such as shear wave imaging or elastography where high frame rates are required' appears to contain a double negative; it likely should read 'unsuitable' or 'less suitable.'
  2. [Table V] The column header of Table V appears to list 'Stage 1' twice; the third column should be labeled 'Stages 1+2' or 'Combined stages' to match the text.
  3. [Section II-E] The power analysis should report the assumed power, alpha, and number of readers in addition to the effect size, so that the stated sample size of 24 image sets (20 after exclusions) can be properly assessed.
  4. [Code Availability] The code availability line contains a space in the URL ('cyclegan verasonics'); it should be a single, clickable URL.
  5. [Section II-A] The criterion for classifying an image as longitudinal or transverse ('the 18 angles most parallel to the muscle fibers') is vague; a concrete angular threshold relative to the fiber direction would improve reproducibility, since the transducer rotation spans 36 angles over 180 degrees.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the two-stage enhancement is trained on external data and evaluated on held-out test sets.

full rationale

The paper's derivation chain is not circular. Stage 1 is a supervised U-Net trained on paired plane-wave inputs and ground-truth compounded/filtered images; its RMSE/SSIM are computed on a held-out test split (Table II), so the match to the ground-truth pipeline is an empirical fit, not a self-fulfilling prediction. Stage 2 is an unpaired CycleGAN trained against an external clinical-domain corpus (Table I), with a held-out test set; the claimed improvements in speckle, CNR, and fiber coherence are measured on test-set ROIs (Table III) and by a reader study. The 'clinical quality' target is defined by the external repositories, not by the model's own outputs, so there is no definitional circularity. Minor self-citations—Paley et al. [28] for acquisition setup and MimickNet [9] as the CycleGAN basis—are contextual and not load-bearing: the cited setup could be replaced by an independent description without affecting the results, and MimickNet is used as architecture, not as evidence for this paper's performance. The Discussion's admissions about hallucination (Section IV) and transverse/longitudinal imbalance (Sections II-B and IV) are important validity limitations—the reader-study 'structural fidelity' is subjective clarity, not verified anatomical correspondence—but they are not circularity. The central enhancement claims are self-contained against external data and held-out evaluation; the minor non-load-bearing self-citations justify a score of 1 rather than 0.

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

The central claim relies on a hand-designed intermediate pipeline, an assumed target clinical style from heterogeneous public data, and subjective reader scores; no new physical entities are introduced.

free parameters (4)
  • CycleGAN loss weights λG and λF = λG=1, λF=0.1
    Tuned via parameter sweeps; balance adversarial, cycle, and identity losses in Eq. (1) and (2).
  • Longitudinal/transverse angle threshold = 18 of 36 angles
    Images at the 18 angles most parallel to muscle fibers are labeled longitudinal, the rest transverse; this arbitrary split affects training labels and the CycleGAN's orientation bias.
  • Plane wave compounding angle set for ground truth = 12 plane waves at -3°, 0°, 3°
    Paired ground truth uses these angles; the U-Net is trained to emulate this specific compounding recipe, so the target quality is tied to this choice.
  • Upsampling resolution = 512×512 from 97×191
    Images are upsampled with bicubic interpolation; resolution is fixed by the U-Net architecture and constrains the output detail.
assumptions (4)
  • domain assumption The public clinical repositories define a single representable 'clinical muscle B-mode' style.
    Section II-B uses four heterogeneous datasets (different scanners, contrasts, orientations) as the CycleGAN target; if no single style exists, the style transfer output is not a valid clinical proxy.
  • domain assumption The PWC plus histogram matching plus unsharp masking pipeline is an appropriate intermediate ground truth.
    Section II-A defines this hand-designed pipeline as the target for the first stage; the final image quality is bounded by this pipeline's fidelity to true clinical appearance.
  • domain assumption Reader ratings of speckle and structural fidelity on a 0-3 Likert scale are valid proxies for clinical image quality.
    Section II-E relies on two neurologists' subjective scores as the primary endpoint; no objective link to diagnostic utility is established.
  • domain assumption The CycleGAN preserves clinically relevant anatomy while changing style.
    Section IV acknowledges the CycleGAN can hallucinate structures; the assumption that improved 'structural fidelity' reflects true anatomy is not verified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Emulating Clinical Quality Muscle B-mode Ultrasound Images from Plane Wave Images Using a Two-Stage Machine Learning Model." pith.science (2026). https://pith.science/paper/XPQGLKQZ

@misc{pith2026241205758,
  author       = {Pith},
  title        = {Pith review of: Emulating Clinical Quality Muscle B-mode Ultrasound Images from Plane Wave Images Using a Two-Stage Machine Learning Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XPQGLKQZ}},
  note         = {Machine review of arXiv:2412.05758}
}
read the original abstract

Research ultrasound scanners such as the Verasonics Vantage often lack the advanced image processing algorithms used by clinical systems. Image quality is even lower in plane wave imaging - often used for shear wave elasticity imaging (SWEI) - which sacrifices spatial resolution for temporal resolution. As a result, delay-and-summed images acquired from SWEI have limited interpretability. In this project, a two-stage machine learning model was trained to enhance single plane wave images of muscle acquired with a Verasonics Vantage system. The first stage of the model consists of a U-Net trained to emulate plane wave compounding, histogram matching, and unsharp masking using paired images. The second stage consists of a CycleGAN trained to emulate clinical muscle B-modes using unpaired images. This two-stage model was implemented on the Verasonics Vantage research ultrasound scanner, and its ability to provide high-speed image formation at a frame rate of 28.5 +/- 0.6 FPS from a single plane wave transmit was demonstrated. A reader study with two physicians demonstrated that these processed images had significantly greater structural fidelity and less speckle than the original plane wave images.

Figures

Figures reproduced from arXiv: 2412.05758 by the authors.

Figure 1
Figure 1. Left column: Single plane wave images taken along the [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. CycleGAN architecture. The first stage U-Net shares the [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 2
Figure 2. Example images from clinical datasets used to train the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Setup used to image a healthy volunteer using the Verasonics [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Example of ROIs used to calculate image metrics. Standard [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 6
Figure 6. Figure 6: Example images taken from the test dataset. The first column [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Images from the test dataset following processing by the first-stage U-Net and the second-stage CycleGAN. Rows 1, 2, and 3 [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 9
Figure 9. Figure 9: Average scores and standard deviations (sample size of 20 [PITH_FULL_IMAGE:figures/full_fig_p006_9.png]
Figure 8
Figure 8. Figure 8: Two sets of paired images evaluated by the readers. Image [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 7
Figure 7. Figure 7: Combining the two stages together, the combined [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 10
Figure 10. Figure 10: Real-time images of the vastus lateralis were acquired [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 42 canonical work pages

  1. [1]

    Ahman, L

    H. Ahman, L. Thompson, A. Swarbrick, and J. Woodward. Under- standing the advanced signal processing technique of real-time adap- tive filters. Journal of Diagnostic Medical Sonography, 25(3):145–160, 2009

    work page 2009

  2. [2]

    Asgariandehkordi, S

    H. Asgariandehkordi, S. Goudarzi, M. Sharifzadeh, A. Basarab, and H. Rivaz. Denoising plane wave ultrasound images using diffusion probabilistic models. IEEE Transactions on Ultrasonics, Ferro- electrics, and Frequency Control , pages 1–1, 2024

    work page 2024

  3. [3]

    Ultrasound Image Enhancement using CycleGAN and Perceptual Loss

    S. Athreya, A. Radhachandran, V . Ivezi ´c, V . Sant, C. W. Arnold, and W. Speier. Ultrasound image enhancement using cyclegan and perceptual loss. arXiv preprint arXiv:2312.11748 , 2023

    work page Pith review arXiv 2023

  4. [4]

    Bottenus, B

    N. Bottenus, B. C. Byram, and D. Hyun. Histogram matching for visual ultrasound image comparison. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 68(5):1487–1495, 2021

    work page 2021

  5. [5]

    Cunningham, M

    R. Cunningham, M. B. S ´anchez, G. May, and I. Loram. Estimating full regional skeletal muscle fibre orientation from b-mode ultrasound images using convolutional, residual, and deconvolutional neural net- works. Journal of Imaging , 4(2):29, 2018

    work page 2018

  6. [6]

    Damerjian, O

    V . Damerjian, O. Tankyevych, N. Souag, and E. Petit. Speckle characterization methods in ultrasound images - a review. IRBM, 35(4):202–213, 2014

    work page 2014

  7. [7]

    F. Faul, E. Erdfelder, A.-G. Lang, and A. Buchner. G* power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39(2):175–191, 2007

    work page 2007

  8. [8]

    J. R. Flynn, S. Ward, J. Abich, and D. Poole. Image quality assessment using the ssim and the just noticeable difference paradigm. In Engi- neering Psychology and Cognitive Ergonomics. Understanding Human Cognition: 10th International Conference , pages 23–30. Springer, 2013

    work page 2013

Show all 44 references
  1. [9]

    Huang, W

    O. Huang, W. Long, N. Bottenus, M. Lerendegui, G. E. Trahey, S. Farsiu, and M. L. Palmeri. Mimicknet, mimicking clinical image post-processing under black-box constraints. IEEE Transactions on Medical Imaging, 39(6):2277–2286, 2020

    2020

  2. [10]

    D. Hyun, A. Wiacek, S. Goudarzi, S. Rothlubbers, A. Asif, K. Eickel, Y . C. Eldar, J. Huang, M. Mischi, H. Rivaz, D. Sinden, R. J. G. van Sloun, H. Strohm, and M. A. L. Bell. Deep learning for ultrasound image formation: Cubdl evaluation framework and open datasets. IEEE Trans...

    2021

  3. [11]

    Isola, J.-Y

    P. Isola, J.-Y . Zhu, T. Zhou, and A. A. Efros. Image-to-image translation with conditional adversarial networks. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 1125–1134, 2017

    2017

  4. [12]

    M. H. Jafari, H. Girgis, N. Van Woudenberg, N. Moulson, C. Luong, A. Fung, S. Balthazaar, J. Jue, M. Tsang, P. Nair, K. Gin, R. Rohling, P. Abolmaesumi, and T. Tsang. Cardiac point-of-care to cart-based ul- trasound translation using constrained cyclegan. International Journal...

    2020

  5. [13]

    Z. Ji, N. Lee, R. Frieske, T. Yu, D. Su, Y . Xu, E. Ishii, Y . J. Bang, A. Madotto, and P. Fung. Survey of hallucination in natural language generation. ACM Computing Surveys , 55(12):1–38, 2023

    2023

  6. [14]

    S. Khan, J. Huh, and J. C. Ye. Pushing the limit of unsuper- vised learning for ultrasound image artifact removal. arXiv preprint arXiv:2006.14773, 2020

    2006 arXiv

  7. [15]

    S. Khan, J. Huh, and J. C. Ye. Contrast and resolution improvement of pocus using self-consistent cyclegan. In Domain Adaptation and Representation Transfer, and Affordable Healthcare and AI for Re- source Diverse Global Health , pages 158–167. Springer International Publishin...

    2021

  8. [16]

    A. E. Knight, C. A. Trutna, N. C. Rouze, L. D. Hobson-Webb, A. Caenen, F. Q. Jin, M. L. Palmeri, and K. R. Nightingale. Full characterization of in vivo muscle as an elastic, incompressible, transversely isotropic material using ultrasonic rotational 3d shear wave elasticity i...

    2022

  9. [17]

    F. W. Kremkau and K. J. Taylor. Artifacts in ultrasound imaging. Journal of Ultrasound in Medicine , 5(4):227–237, 1986

    1986

  10. [18]

    Leitner, R

    C. Leitner, R. Jarolim, B. Englmair, A. Kruse, K. A. L. Hernandez, A. Konrad, E. Y . S. Su, J. Schr ¨ottner, L. A. Kelly, G. A. Lichtwark, M. Tilp, and C. Baumgartner. A human-centered machine-learning approach for muscle-tendon junction tracking in ultrasound images. IEEE Tra...

    1920

  11. [19]

    Z. Li, A. Wiacek, and M. A. L. Bell. Beamforming with deep learning from single plane wave rf data. In2020 IEEE International Ultrasonics Symposium (IUS), pages 1–4, 2020

    2020

  12. [20]

    Liebgott, A

    H. Liebgott, A. Rodriguez-Molares, F. Cervenansky, J. A. Jensen, and O. Bernard. Plane-wave imaging challenge in medical ultrasound. In 2016 IEEE International Ultrasonics Symposium (IUS) , pages 1–4, 2016

    2016

  13. [21]

    J. Long, W. Long, N. Bottenus, and G. Trahey. Coherence-based quan- tification of acoustic clutter sources in medical ultrasound. Journal of the Acoustical Society of America , 148(2):1051, 2020

    2020

  14. [22]

    Y . Lyu, Y . Xu, C. Yang, X. Jiang, X. Zhao, and X. Zhu. Wsrgan: A wavelet-based gan for super-resolution of plane-wave ultrasound images without sampling loss. Engineering Applications of Artificial Intelligence, 127:107384, 2024

    2024

  15. [23]

    X. Mao, Q. Li, H. Xie, R. Y . Lau, Z. Wang, and S. Paul Smolley. Least squares generative adversarial networks. In Proceedings of the IEEE International Conference on Computer Vision, pages 2794–2802, 2017

    2017

  16. [24]

    Marzola, N

    F. Marzola, N. van Alfen, J. Doorduin, and K. M. Meiburger. Deep learning segmentation of transverse musculoskeletal ultrasound images for neuromuscular disease assessment. Computers in Biology and Medicine, 135:104623, 2021

    2021

  17. [25]

    Mittal and S

    S. Mittal and S. Vaishay. A survey of techniques for optimizing deep learning on gpus. Journal of Systems Architecture , 99:101635, 2019

    2019

  18. [26]

    Montaldo, M

    G. Montaldo, M. Tanter, J. Bercoff, N. Benech, and M. Fink. Coherent plane-wave compounding for very high frame rate ultrasonography and transient elastography. IEEE Transactions on Ultrasonics, Ferro- electrics, and Frequency Control , 56(3):489–506, 2009

    2009

  19. [27]

    A. Q. Nichol and P. Dhariwal. Improved denoising diffusion prob- abilistic models. In International Conference on Machine Learning , pages 8162–8171. PMLR, 2021

    2021

  20. [28]

    C. T. Paley, A. E. Knight, F. Q. Jin, S. Moavenzadeh, N. C. Rouze, L. S. Pietrosimone, M. L. Palmeri, and K. R. Nightingale. On the correlation between knee flexion and 3d shear wave speed and amplitude in in vivo vastus lateralis. In 2022 IEEE International Ultrasonics Sympos...

    2022

  21. [29]

    Peetrons

    P. Peetrons. Ultrasound of muscles. European Radiology, 12(1):35–43, 2002

    2002

  22. [30]

    E. D. Pisano, E. B. Cole, B. M. Hemminger, M. J. Yaffe, S. R. Aylward, A. D. A. Maidment, R. E. Johnston, M. B. Williams, L. T. Niklason, E. F. Conant, L. L. Fajardo, D. B. Kopans, M. E. Brown, and S. M. Pizer. Image processing algorithms for digital mammography: A pictorial e...

    2000

  23. [31]

    Ritsche, M

    P. Ritsche, M. V . Franchi, O. Faude, T. Finni, O. Seynnes, and N. J. Cronin. Fully automated analysis of muscle architecture from b- mode ultrasound images with dl track us. Ultrasound in Medicine & Biology, 50(2):258–267, 2024

    2024

  24. [32]

    Rothl ¨ubbers, H

    S. Rothl ¨ubbers, H. Strohm, K. Eickel, J. Jenne, V . Kuhlen, D. Sinden, and M. G ¨unther. Improving image quality of single plane wave ultrasound via deep learning based channel compounding. In 2020 IEEE International Ultrasonics Symposium (IUS) , pages 1–4, 2020

    2020

  25. [33]

    C. Rudin. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature Machine Intelligence, 1(5):206–215, 2019

    2019

  26. [34]

    Saboori and C

    A. Saboori and C. Gr ¨onlund. Unsupervised segmentation and data augmentation in image sequences of skeletal muscle contraction by cycle-consistent generative adversarial network. In 2023 International Conference on Modeling, Simulation & Intelligent Computing (MoSI- Com), pag...

    2023

  27. [35]

    Salimans and J

    T. Salimans and J. Ho. Progressive distillation for fast sampling of dif- fusion models. International Conference on Learning Representations, 2022

    2022

  28. [36]

    Salvi, C

    M. Salvi, C. Caresio, K. M. Meiburger, B. De Santi, F. Molinari, and M. A. Minetto. Transverse muscle ultrasound analysis (trama): Robust and accurate segmentation of muscle cross-sectional area. Ultrasound in Medicine & Biology , 45(3):672–683, 2019

    2019

  29. [37]

    C. A. Trutna, A. E. Knight, F. Q. Jin, N. C. Rouze, L. S. Pietrosimone, A. P. Toth, L. D. Hobson-Webb, M. L. Palmeri, and K. R. Nightingale. Group shear wave speed viscoelastic analysis using 3d rotational volumetric shear wave imaging in relaxed and contracted in vivo muscle....

    2021

  30. [38]

    Usama, E

    M. Usama, E. Nyman, U. Naslund, and C. Gronlund. A domain adaptation model for carotid ultrasound: Image harmonization, noise reduction, and impact on cardiovascular risk markers. arXiv preprint arXiv:2407.05163, 2024

    2024 arXiv

  31. [39]

    H. G. A. van der Pol, L. M. van Karnenbeek, M. Wijkhuizen, F. Geldof, and B. Dashtbozorg. Deep learning for point-of-care ultrasound image quality enhancement: A review. Applied Sciences , 14(16), 2024

    2024

  32. [40]

    T. J. van der Zee and A. D. Kuo. Timtrack: A drift-free algorithm for estimating geometric muscle features from ultrasound images. PLoS One, 17(3):e0265752, 2022

    2022

  33. [41]

    Vi ˜nals and J.-P

    R. Vi ˜nals and J.-P. Thiran. A kl divergence-based loss for in vivo ultrafast ultrasound image enhancement with deep learning. Journal of Imaging, 9(12), 2023

    2023

  34. [42]

    Wijntjes and N

    J. Wijntjes and N. van Alfen. Muscle ultrasound: Present state and future opportunities. Muscle Nerve, 63(4):455–466, 2021

    2021

  35. [43]

    J.-Y . Zhu, T. Park, P. Isola, and A. A. Efros. Unpaired image-to-image translation using cycle-consistent adversarial networks. In Proceedings of the IEEE International Conference on Computer Vision , pages 2223–2232, 2017

    2017

  36. [44]

    ¨Ozbey, O

    M. ¨Ozbey, O. Dalmaz, S. U. H. Dar, H. A. Bedel, S. ¨Ozturk, A. G¨ung¨or, and T. C ¸ ukur. Unsupervised medical image translation with adversarial diffusion models. IEEE Transactions on Medical Imaging, 42(12):3524–3539, 2023

    2023

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.