pith. sign in

arxiv: 2606.02634 · v1 · pith:LIP3PHFZnew · submitted 2026-05-30 · 📡 eess.IV · cs.AI

Echo-POSED: Geometric Self-Distillation for Echocardiography Guidance

Elias Stenhede , Edvart Gr\"uner Bjerke , Joanna Sulkowska , Eivind Bj{\o}rkan Orstad , Ole Jakob Elle , Ulysse C\^ot\'e-Allard , Arian Ranjbar This is my paper

Pith reviewed 2026-06-28 17:52 UTC · model grok-4.3

classification 📡 eess.IV cs.AI
keywords echocardiography guidanceself-supervised learningprobe pose estimationultrasound imaginggeometric equivariance3D volume slicingcardiac phase invariance
0
0 comments X

The pith

Echo-POSED learns probe pose from 2D ultrasound slices of 3D volumes by enforcing motion equivariance and cardiac invariance.

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

The paper introduces Echo-POSED as a self-supervised approach that recommends real-time transthoracic echocardiography probe adjustments directly from single 2D images. It extracts training views by slicing routinely acquired 3D volumes and trains a network to output a representation on SO(3) x SO(3) that changes predictably with simulated probe motion yet stays fixed across heart phases. This removes any requirement for expert view labels or external tracking hardware. If the approach holds, guidance becomes possible on standard 2D machines and across different patients and scanner vendors using only existing 3D datasets for training.

Core claim

Echo-POSED trains on 2D slices from 3D echocardiography volumes to produce a pose representation that remains geometrically consistent under virtual probe perturbations while ignoring cardiac phase, yielding a combined mean angular error of 8.2 degrees between guided and target views in intra-patient simulations that include cardiac motion on held-out and external vendor-shifted datasets.

What carries the argument

Geometric self-distillation that enforces equivariance to probe motions extracted from 3D volumes while maintaining invariance to cardiac phase, producing an SO(3) x SO(3) pose representation from 2D input images.

If this is right

  • The representation maintains geometric consistency when the input view is virtually perturbed.
  • It supports both intra-patient and inter-patient guidance simulations that include cardiac motion.
  • Performance holds across held-out splits and public external datasets that include vendor shifts.

Where Pith is reading between the lines

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

  • The same slicing-and-distillation pattern could be applied to other 3D-to-2D medical imaging tasks where routine volumetric acquisitions exist.
  • Deployment would require checking whether the 8.2-degree error observed in simulation remains acceptable when probe motion is controlled by a human operator in real time.
  • The learned representation might also allow offline simulation of entire probe trajectories for sonographer training.

Load-bearing premise

That 2D slices taken from 3D volumes serve as a realistic stand-in for actual probe movements during live scanning and that the learned representation transfers without further labels or tracking.

What would settle it

A test that measures angular error between the method's recommended probe adjustments and simultaneously tracked ground-truth probe poses during actual live 2D echocardiography scans on patients.

Figures

Figures reproduced from arXiv: 2606.02634 by Arian Ranjbar, Edvart Gr\"uner Bjerke, Eivind Bj{\o}rkan Orstad, Elias Stenhede, Joanna Sulkowska, Ole Jakob Elle, Ulysse C\^ot\'e-Allard.

Figure 1
Figure 1. Figure 1: Left: Echocardiographic 2D-slices are generated from 3D videos, and the model fθ is trained to be equivariant to rotations and translations. Right: The model is fed a target view, it then recommends probe adjustments towards the target view. when specialist expertise is unavailable. Another potential application of such systems is fully autonomous robotic ultrasound systems [3]. Yet most existing guidance … view at source ↗
Figure 2
Figure 2. Figure 2: Echocardiographic acquisition setup and the effect of probe motion on the observed 2D image, with examples of three standard views. The view is determined by probe pose relative to the patient, motivating a geometry-aware formulation of the guidance task. We parametrised the probe pose by two rotations: R (v) for the probe motion on a sphere, and R (p) for its orientation. These are formally introduced in … view at source ↗
Figure 3
Figure 3. Figure 3: Note that the loss only concerns differences between views; [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The ensemble orchestrator model ensures a shared global representation by learning adaptors for each ensemble member. The mean difference in the shared global representation and a known reference is used as a guidance signal for the user. The ensemble disagreement is used to assess image quality and informativeness. 3.2 Representation of rotations and translations We model the probe using two rotations rat… view at source ↗
Figure 5
Figure 5. Figure 5: Model-predicted angle by applied angle across six independent test datasets. N denotes the number of frames in each dataset. A full rotation was applied around the z-axis, whereas rotations from -10◦ to 10◦ are applied to the x and y-axes. as cardiotoxicity monitoring and valvular disease follow-up [21,34]. A necessary step for such reproducibility is the ability to consistently recover a target plane rega… view at source ↗
Figure 6
Figure 6. Figure 6: Selected successful guidance simulations. Echo One Dynamic uses time-varying data, whereas the other datasets are used for navigation in a fixed volume. 6.3 Inter-patient guidance simulation We assess the feasibility of inter-patient guidance using fixed apical two-, three-, and four-chamber target frames selected from a separate patient scan (details in supplementary materials). For each target–volume pai… view at source ↗
Figure 7
Figure 7. Figure 7: Aggregated results for the intra-patient guidance simulations, across held-out and external datasets. The Dynamic scenario results in a similar final actual error, even though the predicted effective target error is larger due to the changing input image. effective pose \protect \hat R_{\mathrm {eff},T} and define geodesic dispersion Deff over the N retained runs as D_{\mathrm {eff}} = \frac {1}{N}\sum _{n… view at source ↗
Figure 8
Figure 8. Figure 8: Guidance effectiveness and final-pose consistency under varying initial probe poses for inter-patient guidance simulation. The results are averaged over three target images representing standard cardiac views. uation would require a 3D dataset with cross-subject view labels or anatomical plane annotations; to our knowledge, such labels are not publicly available. Future work includes extending training bey… view at source ↗
read the original abstract

We introduce Echo-POSED, a self-supervised framework for real-time transthoracic echocardiography (TTE) guidance that recommends probe adjustments directly from 2D ultrasound images, without the need for expert-labelled views or tracked probe trajectories. Instead, it trains on 2D views sliced from routinely acquired 3D echocardiography volumes, enforcing equivariance to probe motions while remaining invariant to cardiac phase, yielding a pose representation on $\mathrm{SO}(3)\times\mathrm{SO}(3)$. Across a held-out split and public external 3D--TTE datasets (including vendor shift), Echo-POSED maintains geometric consistency under virtual perturbations and enables intra- and inter-patient guidance simulations, achieving a combined mean angular error of 8.2 degrees between the guided and target views in intra-patient simulations with cardiac motion.

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 / 1 minor

Summary. The paper introduces Echo-POSED, a self-supervised framework for real-time transthoracic echocardiography (TTE) guidance. It trains on 2D views sliced from 3D echocardiography volumes by enforcing equivariance to probe motions (SO(3)×SO(3) pose representation) while remaining invariant to cardiac phase. The method is evaluated on held-out splits and external vendor-shifted 3D-TTE datasets, reporting a combined mean angular error of 8.2° between guided and target views in intra-patient simulations that include cardiac motion.

Significance. If the learned representation transfers to live clinical scanning, the approach could provide a label-free, tracking-free method for probe adjustment in echocardiography, addressing a practical clinical need. The use of routinely acquired 3D volumes for self-supervision and the reported consistency under virtual perturbations are positive aspects, but the absence of direct validation on physical probe trajectories limits the strength of the significance claim.

major comments (2)
  1. [Abstract] Abstract and evaluation description: the reported 8.2° mean angular error and claims of geometric consistency are obtained exclusively from re-slicing the same 3D volumes used in training (held-out splits and external datasets). No experiments supply tracked real-probe trajectories, acoustic shadowing, or expert-labeled target views on live patients, so the central claim that the representation 'transfers to live clinical scanning' lacks direct supporting evidence.
  2. [Methods] Methods (loss formulation and data handling): the abstract states a numeric result and claims of equivariance/invariance but supplies no equations for the loss, no definition of the SO(3)×SO(3) representation, no data-exclusion rules, and no error-bar details. This prevents verification of whether the metric supports the stated claims about parameter-free or self-supervised behavior.
minor comments (1)
  1. [Abstract] Notation: the pose representation is described as lying on SO(3)×SO(3) without an explicit definition of the two rotation groups or how they map to probe orientation versus cardiac phase.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive feedback on our manuscript. We address the major comments point by point below, clarifying the evaluation design and methodological presentation while acknowledging the scope of our current experiments.

read point-by-point responses

  1. Referee: [Abstract] Abstract and evaluation description: the reported 8.2° mean angular error and claims of geometric consistency are obtained exclusively from re-slicing the same 3D volumes used in training (held-out splits and external datasets). No experiments supply tracked real-probe trajectories, acoustic shadowing, or expert-labeled target views on live patients, so the central claim that the representation 'transfers to live clinical scanning' lacks direct supporting evidence.

    Authors: We agree that all reported results derive from virtual re-slicing of 3D volumes rather than physical probe trajectories acquired on live patients. This design choice enables exact control over SO(3)×SO(3) perturbations and cardiac-phase invariance without additional hardware or annotations, which is central to the self-supervised approach. The 8.2° figure specifically measures intra-patient guidance error under simulated cardiac motion on held-out and external volumes. We will revise the abstract and add a dedicated limitations paragraph to state the evaluation scope more precisely and note that transfer to live scanning remains to be validated in future clinical studies. revision: partial

  2. Referee: [Methods] Methods (loss formulation and data handling): the abstract states a numeric result and claims of equivariance/invariance but supplies no equations for the loss, no definition of the SO(3)×SO(3) representation, no data-exclusion rules, and no error-bar details. This prevents verification of whether the metric supports the stated claims about parameter-free or self-supervised behavior.

    Authors: The abstract provides a concise overview; the full loss (Eq. 3), SO(3)×SO(3) representation (Sec. 3.1), data-exclusion criteria (Sec. 4.1), and error reporting (mean ± std in Tables 1–3) appear in the Methods and Results sections. The framework is self-supervised because it uses only geometric consistency from slicing and requires no expert labels or tracked trajectories. We will add a short cross-reference in the abstract to the relevant sections for improved readability, but we do not believe equations belong in the abstract itself. revision: no

standing simulated objections not resolved
  • Direct experiments with tracked physical probe trajectories, acoustic shadowing, and expert-labeled live-patient views are absent from the study and cannot be supplied without new data collection outside the current 3D-volume dataset.

Circularity Check

0 steps flagged

No significant circularity: self-supervised training on synthetic slices evaluated on held-out data

full rationale

The paper describes a self-supervised framework that enforces equivariance to probe motions (SO(3)×SO(3)) and invariance to cardiac phase on 2D views sliced from 3D echocardiography volumes. The reported mean angular error of 8.2° is measured on held-out splits and external vendor-shifted datasets under virtual perturbations, which are distinct from the training slices. No component of the pose representation is defined in terms of the evaluation metric by construction, no load-bearing self-citations are invoked, and the derivation does not reduce the guidance simulation results to a fitted input or renaming of the input data. This is a standard self-supervised setup whose central claim remains independently testable on the held-out virtual data.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are stated. The approach rests on standard geometric group actions (SO(3)) and the assumption that virtual slicing approximates real probe motion.

pith-pipeline@v0.9.1-grok · 5705 in / 1238 out tokens · 35001 ms · 2026-06-28T17:52:09.744335+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

41 extracted references · 33 canonical work pages · 2 internal anchors

  1. [1]

    In: Linguraru, M.G., Dou, Q., Feragen, A., Giannarou, S., Glocker, B., Lekadir, K., Schnabel, J.A

    Bao, M., Wang, Y., Wei, X., Jia, B., Fan, X., Lu, D., Gu, Y., Cheng, J., Zhang, Y., Wang, C., Zhu, H.: Real-World Visual Navigation for Cardiac Ultrasound View Planning. In: Linguraru, M.G., Dou, Q., Feragen, A., Giannarou, S., Glocker, B., Lekadir, K., Schnabel, J.A. (eds.) Medical Image Computing and Computer As- sisted Intervention – MICCAI 2024. pp. 3...

    work page doi:10.1007/978-3-031-72378-0_30 2024

  2. [2]

    Bernard, O., Heyde, B., Alessandrini, M., Barbosa, D., Camarasu-Pop, S., Cer- venansky, F., Valette, S., Mirea, O., Galli, E., Geleijnse, M.L., Papachristidis, A., Bosch, J.G., D’hooge, J.: Challenge on Endocardial Three-dimensional Ultra- sound Segmentation (CETUS) (2014),https://lirias.kuleuven.be/retrieve/ 9fc21745-b975-4a45-9027-e3285e805b1b

    2014

  3. [3]

    Bi, Y., Jiang, Z., Duelmer, F., Huang, D., Navab, N.: Machine Learning in Robotic Ultrasound Imaging: Challenges and Perspectives. Annual Review of Control, Robotics, and Autonomous Systems7(1), 335–357 (Jul 2024).https://doi.org/ 10.1146/annurev-control-091523-100042,https://www.annualreviews.org/ content/journals/10.1146/annurev-control-091523-100042

    work page doi:10.1146/annurev-control-091523-100042 2024

  4. [4]

    Robots and AI: Illusions and Social Dilemmas

    Carnahan, P., Moore, J., Bainbridge, D., Eskandari, M., Chen, E.C.S., Peters, T.M.: DeepMitral: Fully Automatic 3D Echocardiography Segmentation for Pa- tient Specific Mitral Valve Modelling. In: De Bruijne, M., Cattin, P.C., Cotin, S., Padoy, N., Speidel, S., Zheng, Y., Essert, C. (eds.) Medical Image Computing and Computer Assisted Intervention – MICCAI...

    work page doi:10.1007/978-3-030- 2021

  5. [5]

    Droste, R., Drukker, L., Papageorghiou, A.T., Noble, J.A.: Automatic Probe Movement Guidance for Freehand Obstetric Ultrasound (Jul 2020).https:// doi.org/10.48550/arXiv.2007.04480,http://arxiv.org/abs/2007.04480, arXiv:2007.04480

    work page doi:10.48550/arxiv.2007.04480 2020

  6. [6]

    Frontiers in Cardiovascular Medicine10, 1056055 (Feb 2023)

    Ferraz, S., Coimbra, M., Pedrosa, J.: Assisted probe guidance in cardiac ultra- sound: A review. Frontiers in Cardiovascular Medicine10, 1056055 (Feb 2023). https://doi.org/10.3389/fcvm.2023.1056055,https://www.frontiersin. org/articles/10.3389/fcvm.2023.1056055/full

    work page doi:10.3389/fcvm.2023.1056055 2023

  7. [7]

    Geist, A.R., Frey, J., Zhobro, M., Levina, A., Martius, G.: Learning with 3D ro- tations, a hitchhiker’s guide to SO(3) (Jun 2024).https://doi.org/10.48550/ arXiv.2404.11735,http://arxiv.org/abs/2404.11735, arXiv:2404.11735

    work page arXiv 2024

  8. [8]

    Hagendorff, A., Kandels, J., Metze, M., Tayal, B., Stöbe, S.: Valid and Repro- ducible Quantitative Assessment of Cardiac Volumes by Echocardiography in Patients with Valvular Heart Diseases—Possible or Wishful Thinking? Diagnos- tics13(7), 1359 (Apr 2023).https://doi.org/10.3390/diagnostics13071359, https://www.mdpi.com/2075-4418/13/7/1359

    work page doi:10.3390/diagnostics13071359 2023

  9. [9]

    He, K., Zhang, X., Ren, S., Sun, J.: Deep Residual Learning for Image Recognition (Dec 2015).https://doi.org/10.48550/arXiv.1512.03385,http://arxiv.org/ abs/1512.03385, arXiv:1512.03385

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1512.03385 2015

  10. [10]

    IEEE trans- actions on bio-medical engineering72(7), 2072–2084 (Jul 2025).https://doi

    Huh, J., Klein, P., Funka-Lea, G., Sharma, P., Kapoor, A., Kim, Y.H.: AI-Driven View Guidance System in Intra-Cardiac Echocardiography Imaging. IEEE trans- actions on bio-medical engineering72(7), 2072–2084 (Jul 2025).https://doi. org/10.1109/TBME.2025.3533485

    work page doi:10.1109/tbme.2025.3533485 2072

  11. [11]

    Geometric Self-Distillation for Echocardiography Guidance 17 In: Gomez, A., Khanal, B., King, A., Namburete, A

    Jiang, H., Li, M., Sun, Z., Jia, N., Sun, Y., Luo, S., Song, S., Huang, G.: Structure- aware World Model for Probe Guidance via Large-scale Self-supervised Pre-train. Geometric Self-Distillation for Echocardiography Guidance 17 In: Gomez, A., Khanal, B., King, A., Namburete, A. (eds.) Simplifying Medical Ultrasound. pp. 58–67. Springer Nature Switzerland,...

    work page doi:10.1007/978-3-031-73647-6_6 2025

  12. [12]

    Jordan, K., Jin, Y., Boza, V., You, J., Cesista, F., Newhouse, L., Bernstein, J.: Muon: An optimizer for hidden layers in neural networks (2024),https: //kellerjordan.github.io/posts/muon/

    2024

  13. [13]

    Ultrasound in Medicine & Biology49(9), 1996–2005 (Sep 2023).https://doi.org/10.1016/ j

    Kim, W.J.C., Beqiri, A., Lewandowski, A.J., Mumith, A., Sarwar, R., King, A., Leeson, P., Lamata, P.: Automated Detection of Apical Foreshorten- ing in Echocardiography Using Statistical Shape Modelling. Ultrasound in Medicine & Biology49(9), 1996–2005 (Sep 2023).https://doi.org/10.1016/ j . ultrasmedbio . 2023 . 05 . 003,https : / / www . sciencedirect ....

    1996

  14. [14]

    Kirkpatrick, J.N., Grimm, R., Johri, A.M., Kimura, B.J., Kort, S., Labovitz, A.J., Lanspa, M., Phillip, S., Raza, S., Thorson, K., Turner, J.: Recommendations for Echocardiography Laboratories Participating in Cardiac Point of Care Cardiac Ultrasound (POCUS) and Critical Care Echocardiography Training: Report from the American Society of Echocardiography....

    work page doi:10.1016/j.echo.2020.01 2020

  15. [15]

    Lang, R.M., Badano, L.P., Mor-Avi, V., Afilalo, J., Armstrong, A., Ernande, L., Flachskampf, F.A., Foster, E., Goldstein, S.A., Kuznetsova, T., Lancellotti, P., Muraru, D., Picard, M.H., Rietzschel, E.R., Rudski, L., Spencer, K.T., Tsang, W., Voigt, J.U.: Recommendations for Cardiac Chamber Quantification by Echocardio- graphy in Adults: An Update from th...

    2015

  16. [16]

    Lang,R.M.,Badano,L.P.,Tsang,W.,Adams,D.H.,Agricola,E.,Buck,T.,Faletra, F.F., Franke, A., Hung, J., Pérez De Isla, L., Kamp, O., Kasprzak, J.D., Lancel- lotti, P., Marwick, T.H., McCulloch, M.L., Monaghan, M.J., Nihoyannopoulos, P., Pandian, N.G., Pellikka, P.A., Pepi, M., Roberson, D.A., Shernan, S.K., Shirali, G.S., Sugeng, L., Ten Cate, F.J., Vannan, M....

    work page doi:10.1016/j.echo.2011.11.010 2012

  17. [17]

    Levinson, J., Esteves, C., Chen, K., Snavely, N., Kanazawa, A., Rostamizadeh, A., Makadia,A.:AnAnalysisofSVDforDeepRotationEstimation(Jun2020).https: //doi.org/10.48550/arXiv.2006.14616,http://arxiv.org/abs/2006.14616, arXiv:2006.14616

    work page doi:10.48550/arxiv.2006.14616 2006

  18. [18]

    net/forum?id=BykcN8siz

    Li, Y., Cerrolaza, J.J., Sinclair, M., Hou, B., Alansary, A., Khanal, B., Matthew, J., Kainz, B., Rueckert, D.: Standard Plane Localisation in 3D Fetal Ultrasound Using NetworkwithGeometricandImageLoss.MIDL(Apr2018),https://openreview. net/forum?id=BykcN8siz

  19. [19]

    In: Frangi, A.F., Schnabel, J.A., Davatzikos, C., Alberola-López, C., Fichtinger, G

    Li, Y., Khanal, B., Hou, B., Alansary, A., Cerrolaza, J.J., Sinclair, M., Matthew, J., Gupta, C., Knight, C., Kainz, B., Rueckert, D.: Standard Plane Detection in 3D Fetal Ultrasound Using an Iterative Transformation Network. In: Frangi, A.F., Schnabel, J.A., Davatzikos, C., Alberola-López, C., Fichtinger, G. (eds.) Medical 18 E. Stenhede et al. Image Com...

    2018

  20. [20]

    Springer International Publishing, Cham (2018).https://doi.org/10.1007/ 978-3-030-00928-1_45

    2018

  21. [21]

    Decoupled Weight Decay Regularization

    Loshchilov, I., Hutter, F.: Decoupled Weight Decay Regularization (Jan 2019). https://doi.org/10.48550/arXiv.1711.05101,http://arxiv.org/abs/1711. 05101, arXiv:1711.05101

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1711.05101 2019

  22. [22]

    European Heart Journal 43(41), 4229–4361 (Nov 2022).https://doi.org/10.1093/eurheartj/ehac244, https://academic.oup.com/eurheartj/article/43/41/4229/6673995

    Lyon, A.R., López-Fernández, T., Couch, L.S., Asteggiano, R., Aznar, M.C., Bergler-Klein, J., Boriani, G., Cardinale, D., Cordoba, R., Cosyns, B., Cutter, D.J., De Azambuja, E., De Boer, R.A., Dent, S.F., Farmakis, D., Gevaert, S.A., Gorog, D.A., Herrmann, J., Lenihan, D., Moslehi, J., Moura, B., Salinger, S.S., Stephens, R., Suter, T.M., Szmit, S., Tamar...

    work page doi:10.1093/eurheartj/ehac244 2022

  23. [23]

    The application of machine learning techniques in posttraumatic stress disorder: a systematic review and meta-analysis

    Madani, A., Arnaout, R., Mofrad, M., Arnaout, R.: Fast and accurate view classifi- cation of echocardiograms using deep learning. npj Digital Medicine1(1), 6 (Mar 2018).https://doi.org/10.1038/s41746- 017- 0013- 1,https://www.nature. com/articles/s41746-017-0013-1

    work page doi:10.1038/s41746- 2018

  24. [24]

    https://doi.org/10.48550/arXiv.2408.09931,http://arxiv.org/abs/2408

    Men, Q., Guo, X., Papageorghiou, A.T., Noble, J.A.: Pose-GuideNet: Automatic Scanning Guidance for Fetal Head Ultrasound from Pose Estimation (Aug 2024). https://doi.org/10.48550/arXiv.2408.09931,http://arxiv.org/abs/2408. 09931, arXiv:2408.09931 Geometric Self-Distillation for Echocardiography Guidance 19

    work page doi:10.48550/arxiv.2408.09931 2024

  25. [25]

    A Dynamic Recursive Unified Internet Design (DRUID),

    Men, Q., Zhao, H., Drukker, L., Papageorghiou, A.T., Noble, J.A.: ScanA- head: Simplifying standard plane acquisition of fetal head ultrasound. Medi- cal Image Analysis104, 103614 (Aug 2025).https://doi.org/10.1016/j. media.2025.103614,https://www.sciencedirect.com/science/article/pii/ S1361841525001616

    work page doi:10.1016/j 2025

  26. [26]

    Integration of clinical, pathological, radiological, and transcriptomic data improves prediction for first-line immunotherapy outcome in metastatic non-small cell lung cancer

    Midtvedt, B., Pineda, J., Skärberg, F., Olsén, E., Bachimanchi, H., Wesén, E., Esbjörner, E.K., Selander, E., Höök, F., Midtvedt, D., Volpe, G.: Single-shot self-supervised object detection in microscopy. Nature Communications13(1), 7492 (Dec 2022).https://doi.org/10.1038/s41467- 022- 35004- y,https: //www.nature.com/articles/s41467-022-35004-y

    work page doi:10.1038/s41467- 2022

  27. [27]

    Mitchell, C., Rahko, P.S., Blauwet, L.A., Canaday, B., Finstuen, J.A., Foster, M.C., Horton, K., Ogunyankin, K.O., Palma, R.A., Velazquez, E.J.: Guidelines for Performing a Comprehensive Transthoracic Echocardiographic Examination in Adults: Recommendations from the American Society of Echocardiography. Jour- nal of the American Society of Echocardiograph...

    work page doi:10.1016/j.echo.2018.06.004 2019

  28. [28]

    2011.19.1.1,https://e-jcvi.org/DOIx.php?id=10.4250/jcu.2011.19.1.1

    Nakatani, S.: Left Ventricular Rotation and Twist: Why Should We Learn? Journal of Cardiovascular Ultrasound19(1), 1 (2011).https://doi.org/10.4250/jcu. 2011.19.1.1,https://e-jcvi.org/DOIx.php?id=10.4250/jcu.2011.19.1.1

    work page doi:10.4250/jcu 2011

  29. [29]

    Origami-Inspired Soft Twisting Actuator,

    Narang, A., Bae, R., Hong, H., Thomas, Y., Surette, S., Cadieu, C., Chaudhry, A., Martin, R.P., McCarthy, P.M., Rubenson, D.S., Goldstein, S., Little, S.H., Lang, R.M., Weissman, N.J., Thomas, J.D.: Utility of a Deep-Learning Algo- rithm to Guide Novices to Acquire Echocardiograms for Limited Diagnostic Use. JAMA Cardiology6(6), 624 (Jun 2021).https://doi...

    work page arXiv 2021

  30. [30]

    Ultra- sound in Medicine & Biology49(1), 333–346 (Jan 2023).https://doi.org/10

    Pasdeloup, D., Olaisen, S.H., Østvik, A., Sabo, S., Pettersen, H.N., Holte, E., Grenne, B., Stølen, S.B., Smistad, E., Aase, S.A., Dalen, H., Løvstakken, L.: Real- Time Echocardiography Guidance for Optimized Apical Standard Views. Ultra- sound in Medicine & Biology49(1), 333–346 (Jan 2023).https://doi.org/10. 1016/j.ultrasmedbio.2022.09.006

    2023

  31. [31]

    European Heart Journal

    Sabo, S., Pasdeloup, D., Pettersen, H.N., Smistad, E., Østvik, A., Olaisen, S.H., Stølen, S.B., Grenne, B.L., Holte, E., Lovstakken, L., Dalen, H.: Real-time guid- ance by deep learning of experienced operators to improve the standardization of echocardiographic acquisitions. European Heart Journal. Imaging Methods and Practice1(2), qyad040 (Nov 2023).htt...

    work page doi:10.1093/ehjimp/qyad040 2023

  32. [32]

    Sabo, S., Pettersen, H.N., Smistad, E., Pasdeloup, D., Stølen, S.B., Grenne, B.L., Lovstakken, L., Holte, E., Dalen, H.: Real-time guiding by deep learn- ing during echocardiography to reduce left ventricular foreshortening and mea- surement variability. European Heart Journal - Imaging Methods and Practice 1(1), qyad012 (May 2023).https://doi.org/10.1093...

    work page doi:10.1093/ehjimp/qyad012 2023

  33. [33]

    IEEE transactions on ultrasonics, ferroelectrics, and frequency control 67(12), 2595–2604 (Dec 2020).https://doi.org/10.1109/TUFFC.2020.2981037 20 E

    Smistad, E., Ostvik, A., Salte, I.M., Melichova, D., Nguyen, T.M., Haugaa, K., Brunvand, H., Edvardsen, T., Leclerc, S., Bernard, O., Grenne, B., Lovstakken, L.: Real-Time Automatic Ejection Fraction and Foreshortening Detection Using Deep Learning. IEEE transactions on ultrasonics, ferroelectrics, and frequency control 67(12), 2595–2604 (Dec 2020).https:...

    work page doi:10.1109/tuffc.2020.2981037 2020

  34. [34]

    Medical Image Analysis17(6), 632–648 (Aug 2013).https://doi.org/10.1016/j.media.2013.03.008,https: //www.sciencedirect.com/science/article/pii/S1361841513000388

    Tobon-Gomez, C., De Craene, M., McLeod, K., Tautz, L., Shi, W., Hennemuth, A., Prakosa, A., Wang, H., Carr-White, G., Kapetanakis, S., Lutz, A., Rasche, V., Schaeffter, T., Butakoff, C., Friman, O., Mansi, T., Sermesant, M., Zhuang, X., Ourselin, S., Peitgen, H.O., Pennec, X., Razavi, R., Rueckert, D., Frangi, A.F., Rhode, K.S.: Benchmarking framework for...

    work page doi:10.1016/j.media.2013.03.008 2013

  35. [35]

    EuroIntervention17(14), e1126–e1196 (Feb 2022).https://doi.org/10.4244/EIJ- E- 21- 00009,https: //eurointervention.pcronline.com/doi/10.4244/EIJ-E-21-00009

    Vahanian, A., Beyersdorf, F., Praz, F., Milojevic, M., Baldus, S., Bauersachs, J., Capodanno, D., Conradi, L., De Bonis, M., De Paulis, R., Delgado, V., Freemantle, N., Haugaa, K.H., Jeppsson, A., Jüni, P., Pierard, L., Prendergast, B.P., Sádaba, J.S., Tribouilloy, C., Wojakowski, W.: 2021 ESC/EACTS Guide- lines for the management of valvular heart diseas...

    work page doi:10.4244/eij- 2021

  36. [36]

    Vukadinovic, M., Ieki, H., Sahashi, Y., Ouyang, D., He, B.: Automated Inter- pretable 2D Video Extraction from 3D Echocardiography (Nov 2025).https: //doi.org/10.48550/arXiv.2511.15946,http://arxiv.org/abs/2511.15946, arXiv:2511.15946

    work page doi:10.48550/arxiv.2511.15946 2025

  37. [37]

    Medical Image Analysis70, 101998 (May 2021).https://doi.org/ 10.1016/j.media.2021.101998,https://www.sciencedirect.com/science/ article/pii/S136184152100044X

    Yeung, P.H., Aliasi, M., Papageorghiou, A.T., Haak, M., Xie, W., Namburete, A.I.L.: Learning to map 2D ultrasound images into 3D space with minimal human annotation. Medical Image Analysis70, 101998 (May 2021).https://doi.org/ 10.1016/j.media.2021.101998,https://www.sciencedirect.com/science/ article/pii/S136184152100044X

    work page doi:10.1016/j.media.2021.101998 2021

  38. [38]

    https://doi.org/10.48550/arXiv.2504.13065,http://arxiv.org/abs/2504

    Yue, Y., Wang, Y., Jiang, H., Liu, P., Song, S., Huang, G.: EchoWorld: Learning Motion-Aware World Models for Echocardiography Probe Guidance (Apr 2025). https://doi.org/10.48550/arXiv.2504.13065,http://arxiv.org/abs/2504. 13065, arXiv:2504.13065

    work page doi:10.48550/arxiv.2504.13065 2025

  39. [39]

    Frontiers in Cardiovascular Medicine9, 1016703 (2022).https://doi

    Zhao, D., Ferdian, E., Maso Talou, G.D., Quill, G.M., Gilbert, K., Wang, V.Y., Babarenda Gamage, T.P., Pedrosa, J., D’hooge, J., Sutton, T.M., Lowe, B.S., Legget, M.E., Ruygrok, P.N., Doughty, R.N., Camara, O., Young, A.A., Nash, M.P.: MITEA: A dataset for machine learning segmentation of the left ventricle in 3D echocardiography using subject-specific la...

    work page doi:10.3389/fcvm.2022.1016703 2022

  40. [40]

    Zhou, Y., Barnes, C., Lu, J., Yang, J., Li, H.: On the Continuity of Rotation Representations in Neural Networks (Jun 2020).https://doi.org/10.48550/ arXiv.1812.07035,http://arxiv.org/abs/1812.07035, arXiv:1812.07035

    work page arXiv 2020

  41. [41]

    Ultrasound in Medicine & Biology45(2), 374–384 (Feb 2019).https://doi.org/10.1016/j.ultrasmedbio.2018.07.024

    Østvik, A., Smistad, E., Aase, S.A., Haugen, B.O., Lovstakken, L.: Real-Time Standard View Classification in Transthoracic Echocardiography Using Convolu- tional Neural Networks. Ultrasound in Medicine & Biology45(2), 374–384 (Feb 2019).https://doi.org/10.1016/j.ultrasmedbio.2018.07.024

    work page doi:10.1016/j.ultrasmedbio.2018.07.024 2019