List of Papers Browse by Subject Areas Author List
Abstract
Myocardial point tracking (MPT) has recently emerged as a promising direction for motion estimation in echocardiography, driven by advances in general-purpose point tracking methods. However, myocardial motion fundamentally differs from motion encountered in natural videos, as it arises from physiologically constrained deformation that is spatially and temporally continuous throughout the cardiac cycle. Consequently, motion trajectories typically remain locally confined despite substantial tissue deformation. Motivated by these properties, we revisit the architectural design for MPT and find that coarse initialization in commonly used two-stage coarse-to-fine architectures may be unnecessary in this domain. In this work, we propose a fine-stage-only architecture, EchoTracker2, which enriches pixel-precise features with local spatiotemporal context and integrates them with long-range joint temporal reasoning for robust tracking. Experimental results across in-distribution, out-of-distribution (OOD), and public synthetic datasets show that our model improves position accuracy by 6.5% and reduces median trajectory error by 12.2% relative to a domain-specific state-of-the-art (SOTA) model. Compared to the best general-purpose point tracking method, the improvements are 2.0% and 5.3%, respectively. Moreover, EchoTracker2 shows better agreement with expert-derived global longitudinal strain (GLS) and enhances test-rest reproducibility. Source code available at: https://github.com///.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026/paper/3869_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: https://papers.miccai.org/miccai-2026/supp/3869_supp.zip
Link to the Code Repository
https://github.com/riponazad/ptecho
Link to the Dataset(s)
N/A
BibTex
@InProceedings{AzaMd_EchoTracker2_MICCAI2026,
author = { Azad, Md Abulkalam AND Holmstrøm, Vegard AND Nyberg, John AND Lovstakken, Lasse AND Dalen, Håvard AND Grenne, Bjørnar AND Østvik, Andreas},
title = { { EchoTracker2: Enhancing Myocardial Point Tracking by Modeling Local Motion } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 16889},
month = {September},
page = {pending}
}
Reviews
Review #1
- Please describe the contribution of the paper
The paper proposes EchoTracker2, a myocardial point tracking method for echocardiography that uses the locally constrained nature of cardiac motion. The main contribution is a fine-stage-only architecture that removes the commonly used coarse-to-fine initialization and instead performs correspondence estimation within a confined spatial neighborhood. The method integrates local spatiotemporal feature enrichment (via temporally-aware feature extraction) with joint temporal reasoning across neighboring trajectories. The approach is evaluated on multiple datasets, and shows improved tracking accuracy and agreement with clinically relevant measurements compared to prior methods.
- Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
*A strength is the domain-motivated design, where the method uses the locally constrained and temporally smooth nature of cardiac motion. While the individual components are not entirely novel, their combination results in consistent performance improvements. *Another strength is the comprehensive experimental evaluation. The method is evaluated on multiple datasets, including a large in-house dataset, an ODD setting, and a public dataset, which provides some evidence of generalization. The comparison includes both domain-specific and general-purpose tracking methods, and improvements are consistently observed across datasets. *The paper demonstrates clinical relevance by evaluating downstream (beyond standard tracking) metrics including agreement with expert measurements and reproducibility.
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
*First, the method description lacks clarity and is difficult to follow. Key components of the pipeline are described using heavy notation with limited intuitive explanation. As a result, it is challenging to form a clear understanding of how the method operates in practice. Even though Fig. 2 aims to summarize the approach, it combines multiple stages in a single diagram, which reduces readability. *Second, the central claim that coarse/global motion modeling is unnecessary is not fully supported by the experimental evidence. While Table 3 suggests that removing the initialization stage from prior methods has limited impact, this does not constitute a controlled ablation of the proposed method. Since EchoTracker2 differs from prior work in multiple aspects, it remains unclear to what extent the reported improvements are specifically due to the local-only design, as opposed to other architectural changes. *Third, the ablation studies do not fully isolate the contributions of key components. In Table 2(b), all variants include some form of temporal feature enrichment, and there is no baseline without temporal modeling at the feature level. Similarly, in Table 2(c), all variants incorporate some form of joint reasoning across points, making it difficult to assess the actual contribution of inter-point interactions. *Fourth, the paper somewhat overstates the level of domain-specific modeling. The method is framed as modeling physiologically constrained myocardial motion (e. g. “We design EchoTracker2 to model the locally constrained deformation that characterizes myocardial motion. ”), but in practice this is implemented through architectural design choices such as restricting correspondence to a local spatial neighborhood and removing the coarse stage. While this inductive bias is reasonable, it does not constitute an explicit model, and the framing may therefore be somewhat misleading. *Fifth, the clinical motivation is not clearly articulated. Although the paper mentions downstream applications, it does not clearly explain why accurate myocardial point tracking is critical for clinical decision-making. This makes it harder for readers outside the immediate subfield to appreciate the importance of the problem. *Finally, while the evaluation includes a public dataset (CAMUS), the generalization claims are somewhat limited, as this dataset relies on automatically generated trajectories rather than real-world ground truth. Additional evaluation on more diverse real-world data (e. g. , from different vendors or acquisition settings) would strengthen the conclusions.
- Please rate the clarity and organization of this paper
Poor
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The submission has provided an anonymized link to the source code, dataset, or any other dependencies.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html
N/A
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(3) Weak Reject — marginally below the acceptance threshold, but would not mind if accepted, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The paper proposes a domain-motivated approach for myocardial point tracking in echocardiography, where the method is evaluated on multiple datasets and shows consistent improvements over prior work. However, several issues limit the overall contribution:
*First, the method description lacks clarity, making it difficult to understand how the model operates in practice. Key components are described with heavy notation but limited intuitive explanation, and Fig. 2 does not sufficiently clarify the data flow. This reduces the accessibility of the work. *Second, the central claim that coarse/global motion modeling is unnecessary is not convincingly supported. While the paper presents results suggesting that removing the initialization stage has limited impact for existing methods, this does not constitute a controlled ablation of the proposed architecture. Since multiple components are changed simultaneously, it is unclear to what extent the improvements stem from the local-only design versus other architectural choices. *Third, the novelty is somewhat limited, as the method combines existing building blocks with a domain-specific inductive bias. While this is a reasonable design choice, the work is partly framed as modeling physiologically constrained motion, whereas in practice this is mainly implemented through architectural constraints, rather than explicit modeling.
Overall, while the paper presents a solid approach, the limited novelty, insufficiently supported central claim, and lack of clarity in the method description leads to a weak reject recommendation.
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
N/A
- [Post rebuttal] Please justify your final decision from above.
N/A
Review #2
- Please describe the contribution of the paper
1.The paper shows that coarse-to-fine tracking may be unnecessary for myocardial point tracking in echocardiography. 2.It proposes a local-motion-based tracking framework, EchoTracker2, that focuses on local correspondence and neighboring trajectory reasoning. 3.It demonstrates improved tracking accuracy and better GLS agreement across in-distribution, OOD, and synthetic datasets.
- Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
1.The paper proposes a tracking approach focused only on local motion, which is well suited to the characteristics of echocardiography, and achieves SoTA tracking performance. 2.The paper includes strong comparisons with existing SoTA methods, and the evaluation is well designed for the medical domain, including GLS agreement, test–retest reproducibility, and OOD evaluation. 3.It is particularly interesting that the paper directly shows, through experimental results on existing studies, that the initialization stage provides little performance gain in this domain.
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
Major 1.The ground-truth generation process is not described in enough detail. The paper does not specify which semi-supervised tracking algorithm was used, nor how many clinical experts were involved in the manual verification process. 2.The methodological novelty is somewhat limited, since most of the main components are adapted from existing methods. 3.In Table 3, the AIT for LocoTrack without initialization is reported as N/A, but no explanation is provided.
Minor 1.More details on ethical approval, such as the IRB number or institution, would improve transparency, although I understand that these details may have been omitted at this stage for anonymization purposes.
- Please rate the clarity and organization of this paper
Good
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The submission has provided an anonymized link to the source code, dataset, or any other dependencies.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html
Although it is not a fully matched benchmark for this task, the public EchoNet-LVH dataset provides landmark point annotations, which could still serve as an additional external validation resource and help strengthen the reliability of the evaluation.
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(4) Weak Accept — marginally above the acceptance threshold, but would not mind if rejected, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
Overall, I evaluated the paper positively because the comparison experiments and ablation studies are highly comprehensive and convincingly demonstrate the strong performance of the proposed model. However, I rated it with some reservations due to limitations in the description of the dataset and the relatively modest methodological novelty.
- Reviewer confidence
Very confident (4)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
N/A
- [Post rebuttal] Please justify your final decision from above.
N/A
Review #3
- Please describe the contribution of the paper
- This paper proposed a fine-stage only model for point tracking in echocardiography based on the assumption that unlike generic motion tracking problems, myocardial motion is constrained to a finite region by biomechanics. The authors introduced EchoTracker2, a single-stage dense point tracking model that integrates temporal feature extraction, inter-frame correlation computation via a transformer-based model and temporal refinement. The authors showed that coarse initialization did not significantly imapct tracking performance in the myocardial use case, and showed that their model generally outperformed SOTA models in both in-distribution and out-of-distribution datasets.
- Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
- The paper challenges the assumption that coarse initialization is needed in most motion tracking problems by identifying the unique latent constraints in myocardial motion, and subsequently explicitly showed that coarse initialization is indeed not necessary.
- Methods are generally well-defined and explained
- The result section included multiple proprietary and public datasets and mutliple similar motion tracking models, and demonstrated that the proposed model generally performed the best across almost all datasets. This makes a strong argument in favor of the proposed approach.
- The inclusion of both in-distribution and out-of-distribution results showed good generalizability
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
- This is a rather minor point but I wish the authors could have published an anonymized link to their source code. A github link that’s partially masked is present and I believe that the authors have made the attempt to open source it.
- It might be good to offer more explanation on how exactly are the ground truth landmarks placed, as it seems that the landmark points are spatially very dense given the relatively low resolution of ultrasound images in general. For example, are they interpolated between a few anatomically important landmarks (such as the apex of the left ventricle, the mitral valve root etc).
- Please rate the clarity and organization of this paper
Good
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The authors claimed to release the source code and/or dataset upon acceptance of the submission.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html
N/A
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(5) Accept — should be accepted, independent of rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
- Overall this is an innovative paper that is well written and logically sound, with good results showing both performance improvement and generalizability. My only suggestion for the authors would be to elaborate on the landmark definition since this method is performing dense point tracking.
- Reviewer confidence
Very confident (4)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
N/A
- [Post rebuttal] Please justify your final decision from above.
N/A
Author Feedback
We sincerely thank all reviewers for their time and feedback. We have carefully addressed the major weaknesses as noted below.
Reviewer#1 R1.1 We aim to improve the clarity of the method description and Figure 2 in the camera-ready version. R1.2 By explicitly removing the global initialization stage from SOTA methods and observing minimal performance impact, we believe our results provide sufficient support that global motion modeling has limited benefit in echocardiography. R1.3 For Table 2(b), the benefit of temporal features is well-established, so we focus on comparing different enrichment strategies. For Table 2(c), the advantage of joint reasoning over none is demonstrated in the main results; Table 2(c) specifically compares forms of joint reasoning, showing KNP-Joint is most effective for anatomical deformation. R1.4 We respectfully note that incorporating domain knowledge through architectural design is well-established in deep learning. Our design choices exploit the local constraints of myocardial motion, though not through explicit cardiac motion priors. R1.5 We agree that the clinical motivation can be better emphasised and will strengthen this in the camera-ready version. R1.6 To our knowledge, no public dataset with real-world ground-truth myocardial tracking annotations exists. Our private dataset represents real-world clinical scenarios with high-quality annotations, and we believe our evaluation across both datasets provides sufficient validation within the scope of a MICCAI paper.
Reviewer#2 R2.1 The tracking algorithm is a proprietary GE HealthCare speckle tracking system, so specific details are unavailable. Several cardiologists — including co-authors with many years of experience — were involved in manual verification. We will clarify this in the camera-ready version. R2.2 While components are adapted from existing methods, our contribution lies in their principled selection and problem-specific modification — a process involving extensive experimentation whose effectiveness is reflected in the results. R2.3 Although we deactivated the global stage of LocoTrack, the associated computation still ran internally, making it impossible to report a meaningful AIT. A clarifying note will be added in the camera-ready version. Regardless, removing the global stage is expected to reduce computation time. R2.4 We thank the reviewer for suggesting EchoNet-LVH. While landmark annotations are sparse, they could potentially be adapted to our setting — we will keep this in mind for future work.
Reviewer#4 R4.1 We will ensure a proper public code release is included in the camera-ready version and will keep the anonymized link suggestion in mind for future submissions. R4.2 As described in the dataset section, ground-truth trajectories are obtained by tracking sparse points along the left and right ventricular walls using a semi-automatic proprietary method, verified by clinical experts. We will ensure this is clearly presented in the camera-ready version.
We thank all reviewers once more for their time and constructive engagement. We are confident the camera-ready version will address the raised concerns and further strengthen the paper.
Meta-Review
Meta-review #1
- Your recommendation
Provisional Accept
- Please justify your decision. In case you deviate from the reviewers’ recommendations, explain in detail the reasons why. In case of an invitation for rebuttal, clarify which points are important to address in the rebuttal.
The paper’s core value lies in its domain-inspired simplification of an established paradigm, which is effectively supported by its insightful ablation and comprehensive evaluation. While the method’s description and component-level novelty could be improved, the strength of the central idea and the robustness of the validation are sufficient.