List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Eduardo Castañeda, Masahito Suzuki, Hiroshi Ashikaga, Èric Lluch, Felix Meister, Viorel Mihalef, Chloé Audigier, Andreas Maier, Henry Halperin, Tiziano Passerini
Abstract
Post-ischemic Ventricular Tachycardia (VT) is sustained by a depolarization wave re-entry through channel-like structures within the post-ischemic scar. These structures are usually formed by partially viable tissue, called Border Zone (BZ). Understanding the anatomical and electrical properties of the BZ is crucial to guide ablation therapy to the right targets, reducing the likelihood of VT recurrence. Virtual Heart methods can provide ablation guidance non-invasively, but they have high computational complexity and have shown limited capability to accurately reproduce the specific mechanisms responsible for clinically observed VT. These outstanding challenges undermine the utility of Virtual Hearts for high precision ablation guidance in clinical practice. In this work, fast phenomenological models are developed to efficiently and accurately simulate the re-entrant dynamics of VT as observed in 12-lead ECG. Two porcine models of Myocardial Infarction (MI) are used to generate personalized bi-ventricular models from pre-operative LGE-MRI images. Myocardial conductivity and action potential duration are estimated using sinus rhythm ECG measurements. Multiple hypotheses for the BZ tissue properties are tested, and optimal values are identified. These allow the Virtual Heart model to produce VTs with good agreements with measurements in terms of ECG lead polarity and VT cycle length. Efficient GPU implementation of the cardiac electrophysiology model allows computation of sustained monomorphic VT in times compatible with the clinical workflow.
Link to paper
DOI: https://doi.org/10.1007/978-3-031-43990-2_21
SharedIt: https://rdcu.be/dnwLw
Link to the code repository
N/A
Link to the dataset(s)
N/A
Reviews
Review #1
- Please describe the contribution of the paper
The authors introduce a virtual heart modeling approach to assess re-entry sites in cases of ventricular tachycardia. The method uses LGE images of the ventricle to generate personalized models with scar and border zones. They then optimize the parameters using the electrophysiology data available for each patient.They test their approach on two swine models with VT and evaluate their performance based on error metrics on the estimated and measured cycle lengths and the peaks of the VT signals.
- Please list the main strengths of the paper; you should write about 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 method includes an approach to estimate the subject specific eletrophysiological properties of the tissue. The paper is fair in acknowledging the limitations of the current approach and does provide sufficient validation of their methods.
- Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.
The method contains multiple steps that require user expertise and subjective assessment. Compute times are slow and it is hard to see how, in the current form, this methodology would be extended to the clinic. The authors also fail to cite important literature on the topic
- Please rate the clarity and organization of this paper
Very Good
- Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance
Generally not reproducible. The authors do cite the methods they use to generate their simulations, but those depend on user discretion at the time of assessing fit between each combination of model parameters. There is no description about the MRI parameters to generate the images, nor any example of such.
- Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html
Would be useful to introduce increased automation when determining the best subject-specific parameters. Generally, the evaluation methods and the results should be described in more detail. This includes figures showing examples, how much CL difference varies over computations and, more descriptive captions of the present figures.
- Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making
5
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The present study has limitations in the evaluation presented and the relatively large need of human intervention. However, not many studies of this kind exist and these are useful for reproducibility.
- Reviewer confidence
Very confident
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
N/A
- [Post rebuttal] Please justify your decision
N/A
Review #2
- Please describe the contribution of the paper
This paper presents a validation study of a novel VAAT method based on efficient phenomenological EP models for high-fidelity VT simulation. Via personalization from sinus-rhythm data along with exploring a large parameter space for the BZ configuration, the modeling framework was able to reproduce the indducibility of sustained VT in two animal models, verified by in-vivo ECG data.
- Please list the main strengths of the paper; you should write about 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 presented VAAT method represents a notable improvement over the state-of-the-art in terms of its inclusion of a fast-conduction network, its ability to personalize EP parameters and to match observed ECG of clinical VT, and its substantially-reduced computational cost.
The presented personalization during sinus rhythm for myocardial diffusivity and the clotting time constant of the current gate, and the evidence of the predicted VT morphology matching that of measured ECG has not been done in any prior work to the reviewers’ knowledge and represents a substantial step ahed for state-of-the-arts.
The exploration of the large hypothesis space of the BZ configuration is interesting.
- Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.
The computational efficiency of the presented framework is an important emphasis, but the results did not seem include detailed quantitative numbers comparing the computation cost of the presented framework to those from the state-of-the-art
The results, while promising, are obtained on only two animal models.
The effect of several important modeling assumptions/components could be further discussed (see details below).
- Please rate the clarity and organization of this paper
Very Good
- Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance
It was not clear if the authors intend to make their modeling framework or data open to the community
- Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html
The manuscript can be improved in several aspects.
-
To add exact numbers on the comparison of computational cost to states-of-the-art.
-
Several pacing sites have to be selected in order to simulate sinus-rhythm results. The effect of the placement of these pacing sites on the subsequent parameter estimation should be examined and discussed.
-
The manuscript stated that “Additionally, other points were added in proximity to areas that were visually identified as possible re-entry channels.” It is not clear how these sites will be identified and how they would affect the results. In the two animal models considered, it seems that none of these additional sites yielded sustainable VTs — does this mean they are not helpful, or do the authors believe that the number of sample size is too small to draw this conclusion? Overall, how these “additional” sites can be identified and how they affected the results should be discussed.
-
While the exploration of the BZ hypothesis space is interesting, it also results in a large number of induced sustained VTs. In the two animal models considered, these large number of induced VTs happened to be reduced to a very small number of unique ECG signatures that are consistent with the measured ECG. Do the authors believe that this conclusion can be generalized? What challenges may present if a larger number of unique ECG signatures are induced?
-
- Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making
5
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
This paper, while with moderate innovation in methodology, presents a significant step ahead in the state of the art of VAAT. While the number of sample size (2 animal models) is small, the difficulty and expense in conducting such experiments can be recognized/appreciated. The paper can be improved by adding some missing details regarding computational cost, and adding clarifications on the effect of several key modeling assumptions on the results.
- Reviewer confidence
Confident but not absolutely certain
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
N/A
- [Post rebuttal] Please justify your decision
N/A
Review #3
- Please describe the contribution of the paper
The paper uses personalisation to obtain patient-specific electrophysiological models of porcs. Afterwards, they simulate invasive stimulation of different regions of the myocardium using a biophysical model, which are then compared to the real intervention.
- Please list the main strengths of the paper; you should write about 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 work manages to capture real phenomena by personalising biophysical models, which is difficult.
- Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.
The methodological contributions are limited, and the results are very specific to the electrophysiology use case, with few use of images. I am unsure of the relevance of the work to the general MICCAI community.
- Please rate the clarity and organization of this paper
Very Good
- Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance
The model and datasets are well described, but they provide no code nor raw data.
- Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html
1) A grid search approach to find the optimal parameters might be too inefficient and led to a suboptimal exploration. There are currently other approaches (CMA-ES, Bayesian optimisation…) that will require less evaluations. 2) There are no details on how the synthetic ECG was derived. In particular, it would be important to know if the chest position was accounted for. 3) It would be nice to include also personalisation results, including pseudo ECGs at the sinus rhythm .
Minor comment: 1) Please, check that the abbreviations are correctly capitalised when defined
- Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making
0
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
This is a physical modelling paper, focusing on electrophysiology. I am unsure of whether the general public of MICCAI might be interested by the topic, and I think that this work will be suited for other venue specialised on computational modeling of the heart.
- Reviewer confidence
Confident but not absolutely certain
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
N/A
- [Post rebuttal] Please justify your decision
N/A
Primary Meta-Review
- Please provide your assessment of this work, taking into account all reviews. Summarize the key strengths and weaknesses of the paper and justify your recommendation. 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.
Two reviewers are positive about the method and clinical applications. But they still have concerns about the details of the method and evaluation. Please address the concerns in your rebuttal.
Author Feedback
Dear reviewers,
We would like to express our appreciation for your comments and constructive feedback.
We acknowledge that the deployment of the proposed modeling pipeline in a clinical setting would require additional speedup, automation, and validation. Our work aims at highlighting methodological improvements (accurate anatomical modeling and GPU-based computational methods) which produce a notable improvement over the state of the art and a further step towards translation to the clinics.
Regarding the comments of R1 and R2 about our method’s computational speed, recent studies using similar computational models of electrophysiology have reported considerably inferior performance compared to our approach: 41s of compute time per simulated second in physical time (ours), vs. 1h per simulated second (Prakosa et al. 2018 Nat Biomed Eng), vs. 17min 8s per simulated second, VARP approach (Campos et al. 2022 Medical Image Analysis). In the study by Campos et al, an almost real-time method (VITA) was proposed to detect ablation targets in-silico. However, VITA aims at finding all plausible VT circuits, even the ones that might not clinically manifest VT, potentially resulting in a too extensive ablation. Thus, VITA relies on a simplified electrophysiology model that is not capable of reproducing sustained VT and does not demonstrate fidelity in the simulated ECG traces. Our work focuses instead on identifying the specific circuits and exit sites of clinical relevance, with the goal of proposing the minimal set of ablation targets with the maximal efficacy.
An important finding of our study is that multiple induced sustained VTs are observed as the physical properties of the border zone are modified, but they tend to cluster around a relatively small number of unique ECG signatures. This seems to suggest that a well-defined substrate morphology can only support a limited set of re-entries and VT signatures. Due to the importance of the substrate morphology, we believe that careful consideration must be given to the role of image segmentation and related uncertainty. Within the scope of this work, we focused on careful manual curation of the input data to help reduce the potential impact of uncertainty due to data quality or algorithm performance. As pointed out by all reviewers, implementing automatic and robust algorithms for automatic data analysis is of major importance for the deployment of the system in the clinic. As we scale up the number of cases processed, we plan to incorporate these methods into our pipeline.
Finally, we address R2 comments about our modeling choices for the localization of pacing sites. The animals considered in this study featured a narrow QRS complex in sinus rhythm (70 and 80 ms) despite significant scar burden. Thus, the effect of varying pacing locations for the sinus rhythm simulation was minimal since a relatively high myocardial diffusivity was required to match the measured ECG regardless of the specific activation pattern. We therefore resorted to a standardized selection of pacing locations in sinus rhythm, although personalization strategies can be implemented as an extension of the work and are expected to be significant in particular in presence of long QRS. VT inducibility depends on the pacing point location as well as programmed stimulation protocol: in our study we found that pacing points placed in proximity of a channel-like structure visually identified in the images lead to an increased chance of inducing sustained VT. This allowed us to elicit additional sustained monomorphic VTs, which were not induced with pacing points positioned based on the 17-segment AHA LV model. We hypothesize that pacing near an anatomical channel increases the chances of producing a unidirectional conduction block and self-sustained re-entry. This approach can be automated based on the segmentation of scar and border zone and it is planned as future extension of the work.
Post-rebuttal Meta-Reviews
Meta-review # 1 (Primary)
- Please provide your assessment of the paper taking all information into account, including rebuttal. Highlight the key strengths and weaknesses of the paper, clarify how you reconciled contrasting review comments and scores, indicate if concerns were successfully addressed in the rebuttal, and provide a clear justification of your decision. If you disagree with some of the (meta)reviewer statements, you can indicate so in your meta-review. Please make sure that the authors, program chairs, and the public can understand the reason for your decision.
This work addresses the EP applications and has applications of CAI. I agree with R3 that this is a minority paper of the MICCAI society, but with the novelty and quality of this work, I expect MICCAI is still an interesting venue for this work to be presented and discussed.
Meta-review #2
- Please provide your assessment of the paper taking all information into account, including rebuttal. Highlight the key strengths and weaknesses of the paper, clarify how you reconciled contrasting review comments and scores, indicate if concerns were successfully addressed in the rebuttal, and provide a clear justification of your decision. If you disagree with some of the (meta)reviewer statements, you can indicate so in your meta-review. Please make sure that the authors, program chairs, and the public can understand the reason for your decision.
The limited technical novelty along with small sample size where the approach was evaluated greatly limit enthusiasm regarding the paper. Importantly, the proposed approach seem to depend on several parameters. The sensitivity of the results on those parameters is not examined. Taken everything into account, this paper presents preliminary work that is not ready for publication at MICCAI.
Meta-review #3
- Please provide your assessment of the paper taking all information into account, including rebuttal. Highlight the key strengths and weaknesses of the paper, clarify how you reconciled contrasting review comments and scores, indicate if concerns were successfully addressed in the rebuttal, and provide a clear justification of your decision. If you disagree with some of the (meta)reviewer statements, you can indicate so in your meta-review. Please make sure that the authors, program chairs, and the public can understand the reason for your decision.
While the reviewers are enthusiastic about the clinical utility of the method, the methodological contributions of the paper were limited. There were some concerns about the evaluation of the method and the human intervention, which were not addressed in the rebuttal. While the paper presented a framework that touched upon computational methods with a focus on efficiency, the results were largely qualitative and lacked definite numbers pertaining to the computational cost, especially as the computational efficiency aspect was emphasized. The numbers related to execution time/computational cost were not specified. While the paper made good contributions in outlining specific ways to validate the VAAT study, the structure and relevance of the paper was deemed as out of scope for MICCAI.