Pith. sign in

REVIEW 4 major objections 5 minor 51 references

DUSTrack: Semi-automated point tracking in ultrasound videos

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read DUSTrack claims a fine-tuned per-frame model plus an optical-flow filter can track arbitrary ultrasound points accurately enough to replace task-specific trackers.

desk verdict Useful open-source ultrasound point-tracking toolkit, but the main accuracy claims rest on a ground truth generated with the same optical flow algorithm used in the filter. read the letter →

arxiv 2507.14368 v1 pith:PDL7O3KH submitted 2025-07-18 cs.CV q-bio.QM

Praneeth Namburi , Roger Pallarès-López , Jessica Rosendorf , Duarte Folgado , Brian W. Anthony This is my paper · ORCID
classification cs.CVq-bio.QM
keywords ultrasoundpointtrackingB-modeopticalflowLucas-Kanadejitterreductiontrack-any-pointmodelsmusclefascicleechocardiography
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

DUSTrack is a semi-automated toolkit for tracking arbitrary points in B-mode ultrasound videos. The paper's central claim is that a model fine-tuned on about 25 hand-annotated frames tracks tissue points more accurately than current zero-shot point trackers, and that a bidirectional optical-flow filter removes frame-to-frame jitter without suppressing fast tissue motion. If this is right, one general workflow can replace task-specific methods for measuring cardiac wall motion, muscle deformation, and fascicle length and pennation angle, while also generating high-quality training data for future automated trackers.

What carries the argument

The load-bearing object is the LK-RSTC filter used in a transposed sliding-window configuration: Lucas-Kanade optical flow with reverse sigmoid tracking correction builds short tracklets between anchor estimates, and each frame's final position is the average of all overlapping tracklet estimates passing through that frame. Because the tracklets are short, they inherit Lucas-Kanade's low short-term error; because they are anchored to deep-learning estimates, the average stays globally accurate. The same optical-flow routine also generates interpolated labels for training, augments the training set, and refines the ground-truth annotations.

What would settle it

Have several experts manually label every frame, or use a phantom with known displacements, in the same upper-arm and gastrocnemius videos, then measure DUSTrack's filtered error against that independent ground truth; if the fine-tuned model no longer beats the zero-shot trackers or the filter no longer reduces jitter, the central claim fails.

Watch

Extended reading notes

Core claim

DUSTrack combines per-frame deep-learning localization with a transposed sliding-window optical-flow filter. The deep model sees each frame independently, so errors do not accumulate into drift; the filter generates many short overlapping tracklets with a Lucas-Kanade optical-flow routine that tracks forward and backward with reverse sigmoid correction, then averages the tracklet estimates at each frame to suppress jitter. The paper reports that a ResNet-50 fine-tuned with 25 labeled frames has significantly lower root-mean-square tracking error than four zero-shot TAP models, that the filtered trajectories match the ground-truth power spectrum better than low-pass filters, and that DUSTrack matches a specialized fascicle tracker on fascicle length and pennation angle in a public dataset. It also demonstrates cardiac wall-thickness and chamber-diameter measurements across four cardiac cycles and muscle-deformation patterns during a reaching task.

Load-bearing premise

The accuracy numbers treat DUSTrack-assisted human annotations, which the same optical-flow routine helps interpolate and refine, as unbiased ground truth; if those labels are not independent of the tracker, the reported accuracy and jitter improvements could be partly self-fulfilling.

Editorial extensions

If this is right

  • A single fine-tuned point tracker can yield clinical and biomechanical measurements that today require specialized pipelines: fascicle length, pennation angle, cardiac wall thickness, chamber diameter, and muscle strain.
  • Annotating about 25 frames in one or two motion cycles is enough to reach near-asymptotic accuracy; error drops sharply from 5 to 15 frames and then plateaus.
  • The final LK-RSTC filtering step makes the optional label-augmentation step imperceptible in output trajectories, so augmentation's main remaining value is helping annotators see and verify motion.
  • DUSTrack's measured accuracy of roughly 200 micrometers before filtering sits near the paper's estimated human perceptual threshold of roughly 100 micrometers, giving future zero-shot ultrasound trackers a quantitative target to beat.

Reading between the lines

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

  • Because the same optical-flow routine used in the filter also helps build the ground-truth labels, the accuracy comparison would be stronger with independent manual annotations or a synthetic phantom with known displacements; the reported gap could partly reflect shared algorithm bias.
  • A direct extension would be to fine-tune a video-based point tracker, since the best zero-shot model already showed the lowest jitter but lower accuracy, on DUSTrack-style annotations to combine global accuracy with temporal smoothness; the paper notes hardware costs but does not test this.
  • The transposed tracklet-averaging scheme is a general recipe for any jittery per-frame keypoint estimator, not only ultrasound, whenever local optical flow is reliable enough to bridge short intervals.
  • Clinical validation in patient populations remains untested; pathological tissue or atypical anatomy may violate the assumption that 25 frames in healthy volunteers generalize.
Share X Bluesky LinkedIn Reddit HN

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. The paper introduces DUSTrack, a semi-automated point-tracking toolkit for B-mode ultrasound videos. The workflow combines DeepLabCut-based deep learning models (e.g., ResNet-50) for drift-free point detection with an LK-RSTC optical-flow filter for jitter reduction. The authors claim that a ResNet-50 fine-tuned on about 25 manually annotated frames significantly outperforms zero-shot TAP models (CoTracker3, BootsTAP, LocoTrack, PIPs++) in tracking accuracy, and that the LK-RSTC filter reduces high-frequency noise while preserving rapid tissue motion. The toolkit is demonstrated on cardiac wall motion tracking, upper-arm muscle deformation analysis, and medial gastrocnemius fascicle tracking, with the fascicle comparison performed against an independent public dataset.

Significance. If the central claims hold, DUSTrack would provide a valuable open-source, general-purpose point-tracking framework for ultrasound, with a GUI that supports annotation, augmentation, and iterative model refinement. The fascicle-tracking comparison against an independent public dataset (ground truth from multiple expert raters) is a clear strength, as is the release of the toolkit and code. However, the main accuracy and jitter claims rest on a ground-truth construction that is partly generated by the same LK-RSTC algorithm used in the filter, which, if not resolved, would substantially weaken the reported performance gaps and the filter's evaluation.

major comments (4)
  1. [Methods (annotation step) and Results (first paragraph)] The ground truth for the main accuracy and jitter comparisons is generated with the DUSTrack workflow itself: the Methods state that only approximately 20 of 600 frames are manually annotated and the rest are interpolated with the LK-RSTC algorithm, and the Results describe the ground truth as 'DUSTrack-assisted human annotations ... refined over several iterations and manually corrected.' The same LK-RSTC algorithm is then used as the jitter-reduction filter whose performance is evaluated against this ground truth (Figure 4e-f). This creates a self-referential evaluation: the filtered output is compared to a reference that is itself an LK-RSTC product, so the strong PSD overlap in Figure 4e (the black ground-truth trace is described as 'not visible due to significant overlap with the green trace') is expected and does not by itself validate the filter. Furthermore, the comparison between the fine-tuned ResNet-50 and zero-shot TAP models in Figure 3a may be biased, because zero-shot models are evaluated against a reference that has LK-RSTC-like smoothing characteristics. The manuscript does not report how many of the interpolated frames were manually corrected, nor does it provide any independent validation (e.g., sonomicrometry, synthetic phantoms, or a separate manual-only ground truth). This is load-bearing for the central claims of 'superior accuracy' and 'reduces high-frequency frame-to-frame noise while preserving rapid tissue motion.'
  2. [Results, Figure 4 (LK-RSTC filtering)] The claim that the LK-RSTC filter 'reduces jitter' is supported only by qualitative traces and power spectral density plots (Figure 4e-i); no quantitative RMSE or other numerical error metric is reported comparing the filtered output to the unfiltered model output. Given that the filter is a central contribution, the absence of a quantitative before/after accuracy measure is a significant gap. Reporting, for example, RMSE of filtered versus unfiltered outputs against the ground truth would allow readers to assess the actual improvement.
  3. [Results, Figure 8 (fascicle tracking comparison)] The fascicle tracking comparison against UltraTimTrack is based on a very small sample: the paired t-tests are reported with t3 (d.f. = 3), implying only four participants are included, despite the Methods describing five participants in the public dataset. The claim of 'comparable accuracy' is therefore weakly supported. The paper should clarify the actual number of videos used and consider reporting effect sizes or confidence intervals, or acknowledge the limited statistical power of this comparison.
  4. [Methods, LK-RSTC filtering description] The 'transposed sliding window filter' is described only in prose, without a precise mathematical formulation, pseudocode, or explicit parameter settings (window length, stride, weighting function, etc.). The sliding-window length is listed as a free parameter, but the paper does not report how it was chosen or whether results are sensitive to it. For a methods paper, this hinders reproducibility and makes it difficult for readers to re-implement the filter or assess its robustness.
minor comments (5)
  1. [General] There are several typographical errors: 'speckle-trackig' in the Discussion, a duplicated 'non-invasive' in the Introduction, 'a tracket' in the Figure 4a caption, and 'the a ResNet-50' in the Figure 5a caption. These should be corrected.
  2. [Results, Figure 5] The binomial test result is reported in the text as 'Fig. 5e', but Figure 5e shows the power spectral density analysis, while the preference results appear to be in Figure 5f. Please fix the cross-reference.
  3. [Statistical Analysis and Figure 3a] The Methods state that a Bonferroni correction was applied for multiple testing in the pairwise comparisons between the fine-tuned ResNet-50 and the zero-shot models, but the p-values reported in Figure 3a appear to be uncorrected (e.g., p = 2.71e-16). The paper should clarify whether the reported p-values are adjusted or unadjusted, and if unadjusted, note that all remain significant after correction.
  4. [Discussion] The statement that the fine-tuned ResNet-50 accuracy 'at approximately 200 µm (Fig. 3a)' approaches the perceptual threshold of approximately 100 µm is not directly visible from the figure; please indicate where in the figure these values are shown or provide them in the text.
  5. [Methods, GUI description] The description of the 'transposed sliding window' in the filtering step says that estimates are averaged 'across the same frame,' which is confusing because the window is temporal. Clarify whether the averaging is across overlapping window positions at the same time index or something else.

Circularity Check

1 steps flagged · score 6.0 of 10

Main accuracy and jitter evaluations use a ground truth largely generated by LK-RSTC, the same algorithm used in DUSTrack's filter, making the reported PSD match and fine-tuned-versus-zero-shot gap partly self-referential.

  1. self definitional [Methods (annotation strategy, optical flow refinement); Results (first paragraph of 'State-of-the-art...' and 'LK-RSTC filtering...')]
    "The UI automatically generates intermediate tracking estimates between manually annotated frames using the Lucas-Kanade optical flow algorithm with reverse sigmoid tracking correction (LK-RSTC), reducing manual effort while augmenting the training dataset for deep learning models. ... This step typically produces a few hundred labeled frames, with approximately 20 frames labeled manually (in the span of about 600 sequential frames) and the rest are augmented using the LK-RSTC algorithm. ..."

    The reference for the main quantitative claims is defined through DUSTrack's own pipeline: only ~20/600 frames are manually placed and the intermediate frames are filled by LK-RSTC. The jitter-reduction filter being evaluated is the same algorithm ('the optical flow refinement step employs the LK-RSTC algorithm once again'). Therefore the claim that filtered output PSD 'closely match[es] the ground truth signal' is partly by construction: the reference trajectory and the filtered trajectory share the same LK-RSTC smoothing/interpolation characteristics. The same LK-RSTC-heavy reference is used to rank the fine-tuned ResNet-50 against zero-shot TAP models, so a model trained on DUSTrack labels is structurally favored.

full rationale

The central accuracy and jitter claims are not fully derived from first principles: they are measured against a 'DUSTrack-assisted' ground truth whose non-manual frames are generated with LK-RSTC, the same algorithm used in the filter being evaluated. This makes the PSD overlap in Fig. 4e and part of the RMSE gap in Fig. 3a self-referential. However, the paper is not entirely circular: the fascicle-tracking comparison against UltraTimTrack uses a public dataset with independent expert annotations (Fig. 8), and the UI/modular workflow claims are not affected by this issue. Because the most prominent quantitative claims rest substantially on a self-referential reference, a score of 6 is appropriate; the independent fascicle benchmark and the reported human-correction iterations prevent the score from being higher.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on the prior LK-RSTC algorithm (ref 18) and on the assumption that the DUSTrack-assisted human annotations are valid ground truth. The workflow also assumes that per-frame deep learning removes drift and that short-duration LK tracklets are accurate. The sliding-window length and the recommended number of annotations are hand-chosen design parameters, not fitted to the reported metrics. No new physical entities are introduced.

free parameters (3)
  • LK-RSTC filter sliding window length = 0.6 s (30 frames at 50 Hz)
    Chosen by hand in Methods ("consider a 50 Hz ultrasound video analyzed using a sliding window of 0.6 seconds (30 frames)"). No sensitivity analysis is provided, yet the filter's jitter-vs-motion trade-off depends on this value.
  • Number of manual annotation frames = approximately 25
    Recommended from the accuracy-effort trade-off in Figure 6 ("we recommend annotating approximately 25 frames within 1-2 motion cycles"). The workflow performance depends on this user choice, which is a hand-selected operating point rather than a mathematically derived optimum.
  • Perceptual threshold = approximately 100 µm
    Estimated in Figure 5d as the midpoint between the 90 µm and 140 µm average differences reported by a single blinded evaluator. Used in the Discussion to characterize model accuracy relative to human perception.
assumptions (4)
  • domain assumption LK-RSTC algorithm from Magana-Salgado et al. (2023) provides accurate short-duration tracking in ultrasound videos.
    Invoked throughout the Methods as the basis for annotation interpolation and jitter filtering (Methods, "The algorithm tracks points in both forward and reverse directions..."). The present paper does not re-derive this algorithm.
  • domain assumption DUSTrack-assisted human annotations, refined over several iterations and manually corrected, are unbiased ground truth.
    All quantitative evaluations compare model outputs to these annotations (Results, "The ground truth was DUSTrack-assisted human annotations..."). If these annotations are biased toward LK-RSTC behavior, the reported accuracies are partially self-consistent.
  • domain assumption Processing frames independently prevents accumulation of tracking errors (drift).
    Methods, "they process frames independently without using temporal information, which prevents the accumulation of tracking errors (drift)...". This is a reasonable property of single-frame models, but it is stated without formal or empirical proof in this paper.
  • domain assumption Lucas-Kanade optical flow has low short-term drift and jitter.
    Methods, "This averaging process exploits LK's established low short-term drift and jitter characteristics" (citation 23). The paper relies on this to justify the tracklet filter; no new analysis of LK error characteristics is presented.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DUSTrack: Semi-automated point tracking in ultrasound videos." pith.science (2026). https://pith.science/paper/PDL7O3KH

@misc{pith2026250714368,
  author       = {Pith},
  title        = {Pith review of: DUSTrack: Semi-automated point tracking in ultrasound videos},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PDL7O3KH}},
  note         = {Machine review of arXiv:2507.14368}
}
read the original abstract

Ultrasound technology enables safe, non-invasive imaging of dynamic tissue behavior, making it a valuable tool in medicine, biomechanics, and sports science. However, accurately tracking tissue motion in B-mode ultrasound remains challenging due to speckle noise, low edge contrast, and out-of-plane movement. These challenges complicate the task of tracking anatomical landmarks over time, which is essential for quantifying tissue dynamics in many clinical and research applications. This manuscript introduces DUSTrack (Deep learning and optical flow-based toolkit for UltraSound Tracking), a semi-automated framework for tracking arbitrary points in B-mode ultrasound videos. We combine deep learning with optical flow to deliver high-quality and robust tracking across diverse anatomical structures and motion patterns. The toolkit includes a graphical user interface that streamlines the generation of high-quality training data and supports iterative model refinement. It also implements a novel optical-flow-based filtering technique that reduces high-frequency frame-to-frame noise while preserving rapid tissue motion. DUSTrack demonstrates superior accuracy compared to contemporary zero-shot point trackers and performs on par with specialized methods, establishing its potential as a general and foundational tool for clinical and biomechanical research. We demonstrate DUSTrack's versatility through three use cases: cardiac wall motion tracking in echocardiograms, muscle deformation analysis during reaching tasks, and fascicle tracking during ankle plantarflexion. As an open-source solution, DUSTrack offers a powerful, flexible framework for point tracking to quantify tissue motion from ultrasound videos. DUSTrack is available at https://github.com/praneethnamburi/DUSTrack.

Figures

Figures reproduced from arXiv: 2507.14368 by the authors.

Figure 1
Figure 1. DUSTrack workflow. The process starts with video input and manual annotation of key frames. These annotations can optionally be augmented using the LK -RSTC optical flow algorithm (dashed box). The annotations are used to train a deep learning model, such as the ResNet -50. After model inference, results undergo review. If the results are unsatisfactory, users can make additional manual annotations to refine the mod… view at source ↗
Figure 2
Figure 2. DUSTrack’s graphical user interface. The workflow panel (top left) provides controls for creating DLC projects, training models, and reducing jitter. The video and point annotations panel (right) shows the ultrasound video frame with current point annotations. The selected point is marked here with an arrow. The interface state panel (middle left) displays active variables, including the current annotation layer and… view at source ↗
Figure 3
Figure 3. A finetuned ResNet-50 tracks points in b-mode ultrasound videos with higher accuracy than current state-of-the-art zero-shot methods. (a) Root mean square error (RMSE) in position between the ground truth and the outputs of a finetuned ResNet-50 (trained on 25 labeled frames) and four zero-shot tracking models. The ResNet-50 has significantly lower tracking error compared to CoTracker3 (paired t-test, t35 = -15.56, … view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: The LK-RSTC postprocessing filter in DUSTrack reduces temporal jitter while better preserving both slow and fast motion dynamics compared to low-pass filtering. (a-d) Illustration of the LK-RSTC filtering algorithm, applied to the output of a finetuned ResNet￾50 model …
Figure 5
Figure 5. Figure 5: The LK-RSTC label augmentation step in DUSTrack reduces jitter in initial model outputs but this effect is muted after applying an LK-RSTC filter. (a) The LK-RSTC label augmentation step occurs prior to finetuning the a ResNet-50 model. Power spectral density analysis …
Figure 7
Figure 7. Figure 7: Points tracked via DUSTrack enable extraction of clinically and biomechanically relevant measurements. Applications include cardiac wall motion tracking, muscle deformation tracking and fascicle tracking. (a-c) Using DUSTrack for cardiac wall motion tracking. Long-axis…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 47 canonical work pages

  1. [1]

    & Monteiro, L

    Ferraioli, G. & Monteiro, L. B. S. Ultrasound-based techniques for the diagnosis of liver steatosis. WJG 25, 6053–6062 (2019)

    work page 2019

  2. [2]

    Zügel, M. et al. Fascial tissue research in sports medicine: from molecules to tissue adaptation, injury and diagnostics: consensus statement. Br J Sports Med 52, 1497 (2018)

    work page 2018

  3. [3]

    A., Schumann, B., Townsend, R

    Wong, R. A., Schumann, B., Townsend, R. & Phelps, C. A. A Survey of Therapeutic Ultrasound Use by Physical Therapists Who Are Orthopaedic Certified Specialists. Physical Therapy 87, 986–994 (2007)

    work page 2007

  4. [4]

    & Müller, D

    Schleip, R. & Müller, D. G. Training principles for fascial connective tissues: Scientific foundation and suggested practical applications. Journal of Bodywork and Movement Therapies 17, 103–115 (2013)

    work page 2013

  5. [5]

    D., Wills, B

    Auerbach, J. D., Wills, B. P. D., McIntosh, T. C. & Balderston, R. A. Evaluation of spinal kinematics following lumbar total disc replacement and circumferential fusion using in vivo fluoroscopy. Spine (Phila Pa 1976) 32, 527–536 (2007)

    work page 2007

  6. [6]

    Myers, C. A. et al. In vivo tibiofemoral kinematics during 4 functional tasks of increasing demand using biplane fluoroscopy. Am J Sports Med 40, 170–178 (2012)

    work page 2012

  7. [7]

    O., Wheeler, G

    Kwong, Y., Mel, A. O., Wheeler, G. & Troupis, J. M. Four-dimensional computed tomography (4DCT): A review of the current status and applications. Journal of Medical Imaging and Radiation Oncology 59, 545–554 (2015)

    work page 2015

  8. [8]

    & Chen, G

    Rietzel, E., Pan, T. & Chen, G. T. Y. Four-dimensional computed tomography: Image formation and clinical protocol. Medical Physics 32, 874–889 (2005)

    work page 2005

Show all 51 references
  1. [9]

    Griffiths, R. I. Ultrasound transit time gives direct measurement of muscle fibre length in vivo. Journal of Neuroscience Methods 21, 159–165 (1987)

    1987

  2. [10]

    Taylor, C. R. et al. Magnetomicrometry. Science Robotics 6, eabg0656 (2021)

    2021

  3. [11]

    Haar, G. ter. Ultrasonic imaging: safety considerations. Interface Focus 1, 686–697 (2011)

    2011

  4. [12]

    Harper, C. J. et al. Variability in Diaphragm Motion During Normal Breathing, Assessed With B-Mode Ultrasound. Journal of Orthopaedic & Sports Physical Therapy 43, 927–931 (2013)

    2013

  5. [13]

    A., Bougoulias, K

    Lichtwark, G. A., Bougoulias, K. & Wilson, A. M. Muscle fascicle and series elastic element length changes along the length of the human gastrocnemius during walking and running. Journal of Biomechanics 40, 157–164 (2007)

    2007

  6. [14]

    Michailovich, O. V. & Tannenbaum, A. Despeckling of medical ultrasound images. IEEE Trans. Ultrason., Ferroelect., Freq. Contr. 53, 64–78 (2006)

    2006

  7. [15]

    A., Castro-Ospina, A

    Duarte-Salazar, C. A., Castro-Ospina, A. E., Becerra, M. A. & Delgado-Trejos, E. Speckle Noise Reduction in Ultrasound Images for Improving the Metrological Evaluation of Biomedical Applications: An Overview. IEEE Access 8, 15983–15999 (2020)

    2020

  8. [16]

    & Rohling, R

    Afsham, N., Najafi, M., Abolmaesumi, P. & Rohling, R. A Generalized Correlation-Based Model for Out-of-Plane Motion Estimation in Freehand Ultrasound. IEEE Trans. Med. Imaging 33, 186–199 (2014)

    2014

  9. [18]

    & Anthony, B

    Magana-Salgado, U., Namburi, P., Feigin-Almon, M., Pallares-Lopez, R. & Anthony, B. A comparison of point-tracking algorithms in ultrasound videos from the upper limb. BioMed Eng OnLine 22, 52 (2023)

    2023

  10. [19]

    Weinreb, C. et al. Keypoint-MoSeq: parsing behavior by linking point tracking to pose dynamics. Nat Methods 21, 1329–1339 (2024)

    2024

  11. [20]

    Zee, T. J. van der, Tecchio, P., Hahn, D. & Raiteri, B. J. UltraTimTrack: a Kalman-filter-based algorithm to track muscle fascicles in ultrasound image sequences. PeerJ Comput. Sci. 11, e2636 (2025)

    2025

  12. [21]

    & Ramanan, D

    Vondrick, C. & Ramanan, D. Video Annotation and Tracking with Active Learning

  13. [22]

    & Laporte, C

    Brignol, A., Cheriet, F., Aubin-Fournier, J.-F., Fortin, C. & Laporte, C. Robust unsupervised texture segmentation for motion analysis in ultrasound images. Int J CARS 20, 97–106 (2024)

    2024

  14. [23]

    Lucas, B. D. & Kanade, T. An iterative image registration technique with an application to stereo vision. in IJCAI’81: 7th international joint conference on Artificial intelligence vol. 2 674–679 (1981)

    1981

  15. [24]

    Mathis, A. et al. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci 21, 1281–1289 (2018)

    2018

  16. [25]

    Nath, T. et al. Using DeepLabCut for 3D markerless pose estimation across species and behaviors. Nat Protoc 14, 2152–2176 (2019)

    2019

  17. [26]

    & Sun, J

    He, K., Zhang, X., Ren, S. & Sun, J. Deep Residual Learning for Image Recognition. in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 770–778 (IEEE, Las Vegas, NV, USA, 2016). doi:10.1109/CVPR.2016.90

    2016 doi

  18. [27]

    Howard, A. G. et al. MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications. Preprint at https://doi.org/10.48550/ARXIV.1704.04861 (2017)

    2017 doi arXiv

  19. [28]

    Tan, M. & Le, Q. V. EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks

  20. [30]

    F., Hullfish, T

    Drazan, J. F., Hullfish, T. J. & Baxter, J. R. An automatic fascicle tracking algorithm quantifying gastrocnemius architecture during maximal effort contractions. PeerJ 7, e7120 (2019)

    2019

  21. [31]

    Duffy, G. et al. High-Throughput Precision Phenotyping of Left Ventricular Hypertrophy With Cardiovascular Deep Learning. JAMA Cardiology 7, 386–395 (2022)

    2022

  22. [32]

    Karaev, N. et al. CoTracker3: Simpler and Better Point Tracking by Pseudo-Labelling Real Videos. Preprint at https://doi.org/10.48550/ARXIV.2410.11831 (2024)

    2024 doi arXiv

  23. [33]

    Doersch, C. et al. BootsTAP: Bootstrapped Training for Tracking-Any-Point. in Computer Vision – ACCV 2024 (eds. Cho, M., Laptev, I., Tran, D., Yao, A. & Zha, H.) vol. 15473 483–500 (Springer Nature Singapore, Singapore, 2025)

    2024

  24. [34]

    Cho, S. et al. Local All-Pair Correspondence for Point Tracking. in Computer Vision – ECCV 2024 (eds. Leonardis, A. et al.) vol. 15068 306–325 (Springer Nature Switzerland, Cham, 2025)

    2024

  25. [35]

    W., Shen, B., Wetzstein, G

    Zheng, Y., Harley, A. W., Shen, B., Wetzstein, G. & Guibas, L. J. PointOdyssey: A Large-Scale Synthetic Dataset for Long-Term Point Tracking. in 2023 IEEE/CVF International Conference on Computer Vision (ICCV) 19798–19808 (IEEE, Paris, France, 2023). doi:10.1109/ICCV51070.2023.01818

    2023

  26. [36]

    Lee, G. Y. et al. Cardiac Amyloidosis Without Increased Left Ventricular Wall Thickness. Mayo Clinic Proceedings 89, 781–789 (2014)

    2014

  27. [37]

    Elliott, P. M. et al. 2014 ESC Guidelines on diagnosis and management of hypertrophic cardiomyopathy: The Task Force for the Diagnosis and Management of Hypertrophic Cardiomyopathy of the European Society of Cardiology (ESC). European Heart Journal 35, 2733– 2779 (2014)

    2014

  28. [38]

    Namburi, P. et al. Efficient elastic tissue motions indicate general motor skill. 2025.02.25.636457 Preprint at https://doi.org/10.1101/2025.02.25.636457 (2025)

    2025 doi

  29. [39]

    & Yeo, S.-H

    Verheul, J. & Yeo, S.-H. A Hybrid Method for Ultrasound-Based Tracking of Skeletal Muscle Architecture. IEEE Transactions on Biomedical Engineering 70, 1114–1124 (2023)

    2023

  30. [40]

    Ritsche, P. et al. Fully Automated Analysis of Muscle Architecture from B-Mode Ultrasound Images with DL_Track_US. Ultrasound in Medicine & Biology 50, 258–267 (2024)

    2024

  31. [41]

    E., Allison, J

    Trahey, G. E., Allison, J. W. & von Ramm, O. T. Angle Independent Ultrasonic Detection of Blood Flow. IEEE Transactions on Biomedical Engineering BME-34, 965–967 (1987)

    1987

  32. [42]

    Geyer, H. et al. Assessment of Myocardial Mechanics Using Speckle Tracking Echocardiography: Fundamentals and Clinical Applications. Journal of the American Society of Echocardiography 23, 351–369 (2010)

    2010

  33. [43]

    Kuder, I. M. et al. An Optimization Approach for Creating Application-specific Ultrasound Speckle Tracking Algorithms. Ultrasound in Medicine and Biology 50, 1108–1121 (2024)

    2024

  34. [44]

    Amundsen, B. H. et al. Noninvasive myocardial strain measurement by speckle tracking echocardiography: validation against sonomicrometry and tagged magnetic resonance imaging. J Am Coll Cardiol 47, 789–793 (2006)

    2006

  35. [45]

    Azad, M. A. et al. EchoTracker: Advancing Myocardial Point Tracking in Echocardiography. in Medical Image Computing and Computer Assisted Intervention – MICCAI 2024 (eds. Linguraru, M. G. et al.) vol. 15004 645–655 (Springer Nature Switzerland, Cham, 2024)

    2024

  36. [46]

    & Salcudean, S

    Chen, W., Schmidt, A., Prisman, E. & Salcudean, S. E. PIPsUS: Self-supervised Point Tracking in Ultrasound. in Simplifying Medical Ultrasound (eds. Gomez, A., Khanal, B., King, A. & Namburete, A.) vol. 15186 47–57 (Springer Nature Switzerland, Cham, 2025)

    2025

  37. [47]

    H., Memariani, A

    Jarrahi, M. H., Memariani, A. & Guha, S. The Principles of Data-Centric AI. Commun. ACM 66, 84–92 (2023)

    2023

  38. [48]

    Zha, D. et al. Data-centric Artificial Intelligence: A Survey. ACM Comput. Surv. 57, 1–42 (2025)

    2025

  39. [49]

    & Tourneret, J.-Y

    Ouzir, N., Basarab, A., Lairez, O. & Tourneret, J.-Y. Robust Optical Flow Estimation in Cardiac Ultrasound Images Using a Sparse Representation. IEEE Transactions on Medical Imaging 38, 741– 752 (2019)

    2019

  40. [50]

    Mitchell, C. et al. Guidelines for Performing a Comprehensive Transthoracic Echocardiographic Examination in Adults: Recommendations from the American Society of Echocardiography. Journal of the American Society of Echocardiography 32, 1–64 (2019)

    2019

  41. [51]

    J., Bonow, R

    Spirito, P., Maron, B. J., Bonow, R. O. & Epstein, S. E. Occurrence and significance of progressive left ventricular wall thinning and relative cavity dilatation in hypertrophic cardiomyopathy. The American Journal of Cardiology 60, 123–129 (1987)

    1987

  42. [52]

    Caenen, A. et al. Continuous shear wave measurements for dynamic cardiac stiffness evaluation in pigs. Sci Rep 13, 17660 (2023)

    2023

  43. [53]

    layers" simultaneously—in this example, the

    Hardy, E. J. et al. Exploring the utility of ultrasound to assess disuse atrophy in different muscles of the lower leg. J Cachexia Sarcopenia Muscle 15, 2487–2496 (2024). Figures Figure 1: DUSTrack workflow. The process starts with video input and manual annotation of key fram...

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.