List of Papers Browse by Subject Areas Author List
Abstract
3D ultrasound (US) imaging has shown significant benefits in enhancing the outcomes of percutaneous liver tumour ablation. Its clinical integration is crucial for transitioning 3D US into the therapeutic domain. However, challenges of tumour identification in US images continue to hinder its broader adoption. In this work, we propose a novel framework for integrating 3D US into the standard ablation workflow. We present a key component, a clinically viable 2D US–CT/MRI registration approach, leveraging 3D US as an intermediary to reduce registration complexity. To facilitate efficient verification of the registration workflow, we also propose an intuitive multimodal image visualization technique. In our study, 2D US–CT/MRI registration achieved a landmark distance error of ~2–4 mm with a runtime of 0.22 s per image pair. Additionally, non-rigid registration reduced the mean alignment error by ~40% compared to rigid registration. Results demonstrated the efficacy of the US–CT/MRI registration to improve tumour identification. Our novel 3D US integration framework improved ablation by enhancing tumour visibility, multimodal visualization and tumour coverage, highlighting its potential to expand the therapeutic role of 3D US in other applications.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2025/paper/4548_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to the Code Repository
https://github.com/Xingorno/MICCAI2025-2DUS-to-CTMRI-Multimodal-Registration
Link to the Dataset(s)
N/A
BibTex
@InProceedings{XinShu_ANovel_MICCAI2025,
author = { Xing, Shuwei AND Cool, Derek W. AND Tessier, David AND Chen, Elvis C. S. AND Peters, Terry M. AND Fenster, Aaron},
title = { { A Novel Framework for Integrating 3D Ultrasound into Percutaneous Liver Tumour Ablation } },
booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025},
year = {2025},
publisher = {Springer Nature Switzerland},
volume = {LNCS 15968},
month = {September},
page = {24 -- 34}
}
Reviews
Review #1
- Please describe the contribution of the paper
The paper describes a framework incorporating 3D ultrasound scanning as an intermediate step for 2D ultrasound guided percutaneous liver ablation.
- 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 is clearly written and sufficiently describes the methods.
- 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.
- The core of the paper is the 2D-3D registration algorithm, but this is previous work of the authors (withheld due to anonymity). The rest of the work, despite its practical value for some settings, lacks sufficient novelty.
- The proposed setup seems rather cumbersome, and not compatible with existing 3D US probes. Its potential for routine integration in clinical workflows is not convincingly addressed.
- Experiments are performed on few cases and do not sufficiently cover a range of variability that could be sufficient to prove the performance and robustness of the method.
- Please rate the clarity and organization of this paper
Satisfactory
- 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.
- 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/2025/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.
(2) Reject — should be rejected, independent of rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The lack of novelty of the work, in addition to limitations in the methodology and experiments, lead me to rejection.
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
Reject
- [Post rebuttal] Please justify your final decision from above.
Even after the rebuttal, I still consider that the limited novelty and the limitations of the reported setup are of importance, and thus recommend rejection.
Review #2
- Please describe the contribution of the paper
In this work, the authors integrate existing approaches of liver vessel segmentation (nnUnet), multi-modal image registration (first 3D CT or MRI to a 3D US volume, then 2D slices to 2D US), visualization (blended MPRs and volume rendering) and guidance into a navigation system for liver ablations.
- 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) Demonstration of clinical feasibility of 3D ultrasound for ablation guidance: Great to see that this method reached the clinical threshold of 5 mm in registration accuracy.
- 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.
1) Rather incremental contribution. This is my major concern with this paper. Multi-modal image registration is not new, as the authors themselves have to point out. It has been done in 2D and 3D [10], and already on other anatomies such as prostate [29]. There are medical devices on the market achieving robotic 3D US-to-CT/MRI registration on the liver with good accuracy (e.g. Histosonics), and somewhat recently, sufficiently fast, direct multi-modal registration has been achieved (Ronchetti et al, MICCI 2013, https://link.springer.com/chapter/10.1007/978-3-031-43999-5_72). Ulrasound calibration is - although often tricky to get right - a solved problem. Mluti-planar reconstructions are an industry standard, also both Coherent Point Drift and TPS-based registration have been proposed years ago. In addition, the authors cite three of their own works anonymously, so it hard to judge if this manuscript offers substantial improvements to warrant publication at MICCAI.
2) Missing details of novel aspects. The paper unfortunately remains superficial when it comes to their really novel contributions: How does the MVR part different from the state-of-art? How were the transfer functions etc. chosen? How is TPS parametrized, how are the control points chosen? Is TPS suited to model realistic deformation in this use-case? What images is the algorithm applied in the first place? 3D US in breath-hold?
- 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 does not provide sufficient information for reproducibility.
- 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/2025/en/REVIEWER-GUIDELINES.html
1) Calibration a) The calibration approach looks a bit extensive for a robot with 2 dofs. Could you just estimate distance between the center of rotation and the translation axis? Since absolute pose is not necessary, an image-based calibration technique could optimize all parameters at once. I believe this is implemented in Slicer. b) The calibration errors are quite high, basically spanning almost the entire accuracy threshold of 5 mm. Do you know what’s the most contributing factor here?
2) 3D-3D registration: Some details on the characteristics of the images and how they were annotated, might be useful to the reader. Why was CPD used, and not an image-registration algorithm?
3) Since you already have an NDI camera, you could validate the 2D-3D registration alone with just the tracking data. This would make it possible to assess the accuracy without the bias from that specific motorized 3D probe calibration.
- 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.
(2) Reject — should be rejected, independent of rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The major weaknesses require a substantial revision of this paper to increase clarity, provide enough details for reproduction, and to properly delineate this paper’s contribution from the authors’ own prior work.
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
Reject
- [Post rebuttal] Please justify your final decision from above.
Major concerns regarding the novelty and incremental contribution with respect to the authors’ own work remain.
Review #3
- Please describe the contribution of the paper
This paper contributes a registration approach that makes it feasible to incorporate 3D US into the percutaneous liver ablation workflow. To achieve this, 3D US is used as an intermediate step for 2D US to CT/ MRI image registration. A visualization strategy for intra-procedural alignment is also presented. The contribution is evaluated on a dataset consisting of 95 images from 16 volunteers and 13 patients. The registration error, inference time, and visualization advantages are presented and discussed.
- 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 has a very nice and extensive literature review
- The dataset consists of a large number of images and user variability (patient and healthy volunteer)
- The figures are great
- Overall, I think this work has very high potential to be impactful in liver ablation and the contribution is thoroughly evaluated and well written
- 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.
- As outlined in the introduction, it seems that the prior work has used similar methodology to align 2D US with MRI for prostate imaging. I understand that this new clinical domain makes this task unique and a totally new application but I’d be interested in a clear comparison or section of the paper that demonstrates how the core contributions of that work differ from this one and any domain specific changes that were essential to liver.
- In section 2.1, I’m not completely clear on what the dynamic reference body in the phantom is or why you need the NDI tracker when the probe is fixed to the mechatronic arm - I assume this is your ground truth. In Figure 1 maybe showing the tracker would help? or a photo of the real phantom if that’s where it is used.
- If the data is consistent between the steps, a subsection on the data (what is included) would help rather than mentioning it briefly in the Rigid US-CT section. A few sentences throughout the paper reference specific details so having it all collected somewhere is useful for referring back.
- related to my previous point when you get into section 3.2, there are more details about the data that I think would be better organized in a collective data section
- I’m a bit unclear on the ground truth - for the 3D US - CT/MRI the centerline distance seems to be used for evaluation. A visual would help clarify this. Overall, some very clear, high level explanation of how the results are obtained might help the paper read easier.
- What do the white arrows in Figure 5d show? Am i missing that explanation?
- The clinical workflow for intra-procedural use of this technology is not completely clear to me, a bit of a higher level description explaining how the clinician would interact with this or the imagined workflow would help - specifically, the manual adjustment or confirmation in section b of the workflow and the intention for use in the procedure (would the robot be there while the clinician does the ablation?) Figure 3 is close but I don’t immediately get the whole vision
- 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.
- 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/2025/en/REVIEWER-GUIDELINES.html
I think my comments above should be sufficient.
- 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?
I do not feel that any of the weaknesses I have outlined are significant enough to preclude this paper from inclusion in the MICCAI proceedings. Overall, I think this is a very important contribution for guidance in percutaneous liver ablation.
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
Accept
- [Post rebuttal] Please justify your final decision from above.
In the revised manuscript, I think the authors should clarify the details that are included in their feedback (E specifically). Overall, I feel confident with my initial assessment, and after reading the response to the novelty comments from the other reviewers, I agree with the authors that the demonstration of a complete, clinically validated system as shown here is a good fit for MICCAI and is sufficiently novel.
Author Feedback
A. Novelty and contributions (R1, R2). The primary contribution of our study is the development of a complete, clinically validated 3D US liver ablation guidance system that enables practical 2D US–CT/MRI registration, improving treatment effectiveness and accessibility. Our CAI paper aligns with the MICCAI scope, which supports application studies demonstrating clinical value of existing techniques (see MICCAI guidelines). We support this by developing and integrating hardware and software components (e.g., mechatronic arm, 3D US scanner, registration, and visualization) into an end-to-end intervention system. As R3 noted, this represents significant clinical advancement. To our knowledge, this is the first study to systematically demonstrate a 3D US-assisted registration framework for liver therapy. While more human cases could strengthen the study (R1), system deployment in the operating room and its evaluation by physicians already mark a significant step toward clinical translation.
B. Clinical compatibility and workflow (R1, R2, R3). 1) Our system is positioned beside the surgical bed, opposite the physician, and remains in the room throughout the procedure (R3). It is fully compatible with standard workflows and commercial 2D US probes, as validated in our 32-patient trial [1]. While also compatible with 3D US probes (R1), 2D US is routinely used in procedures, and reliance on costly 3D probes would limit accessibility, particularly in underserved settings. 2) 3D US images are acquired during a breath-hold (R2), followed by physician review for tumour measurement and ablation planning. Registration with CT/MRI is then verified via MVR visualization, with manual correction through control-point placement and adjustment in our software if needed (see repository). Needle insertion and ablation proceed under 2D US, enhanced by the aligned CT/MRI. Notably, manual correction (step b) adds minimal time beyond standard visual alignment checks.
C. Method comparisons (R2, R3). 1) Unlike the prostate, liver interventions face unique challenges—respiratory motion and limited anatomical correspondence between 3D US and CT/MRI—not addressed by methods for motion-stable prostates. 2) Our system presents a complete, clinically validated registration pipeline. Both 3D US–CT/MRI and 2D–3D US registration steps are essential to meet clinical demands for accuracy, runtime and robustness. However, Ronchetti et al. and Histosonics address only 3D US–CT/MRI registration without end-to-end integration (R2). Ronchetti et al. achieved <25 mm registration accuracy in only 75.5% of cases, far from the clinical 5 mm threshold. For Histosonics, we could not find publicly disclosed registration framework and accuracy for comparison.
D. Reproducibility (R2). 1) Our MVR is novel in its integration as a software tool and human validation. Fig. 2b shows its effectiveness compared to others; implementation details are summarized (para 2, p. 5). 2) The transfer function was manually tuned by adjusting scalar opacity mapping to enhance liver structures (e.g., vessels) while suppressing intravascular blood, improving alignment visibility (para 2, p. 5). 3) To constrain the ROI for TPS, 50 boundary control points are sampled via spherical Fibonacci sampling on the intersection of nnUNet-segmented liver (CT/MRI) and physician-selected spherical ROI (3D US). Moving control points are then placed based on observed misalignments, enabling smooth, continuous deformation, which is widely used.
E. Clarification (R2). 1) A 6-DoF model is required in our system. The calibration error (2.79 mm) stems from arm calibration (~2 mm), probe calibration, and minor manufacturing imperfections. Tracking is used only for registration initialization; image-based optimization (step c) ensures final alignment accuracy. 2) While effective for tracking external objects, NDI camera cannot capture internal liver motion and was therefore not used for accuracy assessment.
Meta-Review
Meta-review #1
- Your recommendation
Invite for Rebuttal
- If your recommendation is “Provisional Reject”, then summarize the factors that went into this decision. In case you deviate from the reviewers’ recommendations, explain in detail the reasons why. You do not need to provide a justification for a recommendation of “Provisional Accept” or “Invite for Rebuttal”.
N/A
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Reject
- Please justify your recommendation. You may optionally write justifications for ‘accepts’, but are expected to write a justification for ‘rejects’
N/A
Meta-review #2
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Accept
- Please justify your recommendation. You may optionally write justifications for ‘accepts’, but are expected to write a justification for ‘rejects’
I will align with the accept decision for this work due to its significant clinical advancement in developing a complete, clinically validated 3D ultrasound liver ablation guidance system, which integrates various components into a practical end-to-end solution evaluated in a patient trial. The authors convincingly argue in their rebuttal for the novelty of their systematic application and adaptation of registration techniques specifically for the challenges of liver therapy, differentiating it from prior work on other organs. Furthermore, their thorough rebuttal effectively addressed reviewer concerns regarding clinical workflow, methodological comparisons, and reproducibility, demonstrating the work’s value and readiness for dissemination. Especially Section E in the rebuttal is very important.
Meta-review #3
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Accept
- Please justify your recommendation. You may optionally write justifications for ‘accepts’, but are expected to write a justification for ‘rejects’
N/A