List of Papers Browse by Subject Areas Author List
Abstract
A digital twin (DT) is a dynamic virtual model that mirrors a physical system, with promising applications in surgical planning, guidance, and outcome assessment. While DTs can represent various key aspects of surgery, such as patient anatomy and surgical tools, implants remain difficult to integrate due to tracking challenges related to occlusions by soft tissue and their small size. Consequently, current surgical DTs lack implant integration, a critical limitation in trauma surgery. To address this challenge, this work presents an automated method to integrate surgical implants-plates and screws-into DTs during bone fracture platings. The solution leverages surgical tracking data to analyze interactions between surgical tools and patient anatomy. By combining deterministic algorithms with a machine learning-based activity classification model, DTs of implants can be reconstructed without requiring direct tracking. A study involving 28 participants-5 medical students, 12 residents, and 11 attending physicians-evaluated detection reliability and geometric accuracy on a comminuted ulnar fracture. Results showed a screw detection rate of 96.4% and a plate detection rate of 100% across 112 screws and 28 plates. Screw and plate placement had Root Mean Square Errors of 1.52 mm and 0.94 mm respectively-comparable to or better than existing surgical DTs. These findings confirm the feasibility of dynamic implant integration, marking a significant step toward comprehensive DT solutions for trauma surgery. This advancement has the potential to enhance intraoperative visualization and postoperative assessment, ultimately improving patient care.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2025/paper/4185_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to the Code Repository
N/A
Link to the Dataset(s)
N/A
BibTex
@InProceedings{StaTob_Automated_MICCAI2025,
author = { Stauffer, Tobias and Reber, Manuel and Fellmann, Léon and Babst, Reto and Meboldt, Mirko and Lohmeyer, Quentin},
title = { { Automated Integration of Surgical Implants into Digital Twins for Trauma Surgery } },
booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025},
year = {2025},
publisher = {Springer Nature Switzerland},
volume = {LNCS 15968},
month = {September},
}
Reviews
Review #1
- Please describe the contribution of the paper
This paper proposes a process to create digital twin for implants, particularly plates and screws, for bone fracture plating. The system uses a combination of 6D tool tracking and machine learning to mimic implants placed into a real phantom into a digital twin. The contribution of the paper would be clearer if it clarified why having implants in the digital twin is important and how it could improve patient care.
- 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 goal of the method was clearly stated
- Large user study group with varied levels of expertise
- Clear experimental setup
- 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 paper presents a very set sequence of three screws, and then a plate. Unclear how the system would generalize to any other implants
- Although users varied in skill level, there was no analysis based on different user groups. This is particularly of interest since presumably the same 11 experts provided the training data for the neural network
- Unclear how the performance in phantoms would translate to clinical utility
- 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 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
- Intro’s claim of evaluation on 28 surgeons seems overstated since the experimental setup claims some of the 28 are trainees.
- Missing word “The tools required for the procedure were placed on a separate table next to the setup, as shown Fig. 3b.”
- Fig. 1 would benefit from more description and context. It’s hard to interpret right now.
- Fig. 4 would benefit from an axis of to give a sense of scale
- Images of the custom screw marker would aid reproducibility.
- How similar is the deformation experienced by the phantoms compared to human bones?
- Occlusion seems to be a problem even during the simplified phantom setup. ORs are pretty busy - how could the system be adapted to avoid line-of-sight issues?
- 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 — could be rejected, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
While the application of tracking technology to digital twin seems novel, the underlying methods are not. This would be fine if the paper presented sufficient clinical motivation for why this is an important problem to solve, if the translation efforts seemed generalizable to other classes of problems, or some other reason why this paper makes a significant contribution. I recommend weak reject since additional information may clarify the significance of the paper.
- 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.
The rebuttal letter does not make a strong case for the clinical translatability and significance. The fundamental issue is still occlusion, which the rebuttal letter confirms is not addressed by this work. The novelty of the proposed work is unclear.
Review #2
- Please describe the contribution of the paper
This paper proposes an indirect method of creating a digital representation of surgical implants for analysis of implant stability, without the need for intra-/postoperative radiation based imaging. The method entailed tracking of the tools using a navigation system (Atracsys Fusion Track 250), relaying information to move their virtualized counterparts in Unity, in order to estimate drill holes, placement and the length of the screws by tracking the screwdriver and identifying screwing action with LSTM. Finally the position and orientation of the plate is estimated based on the positions of the screws. Experiments were done with 28 participants, with RMSE of 1.52mm for the screws estimation, and 0.94mm for the plate estimation.
- 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.
High registration accuracy of the digital twin to the real counter part, as compared to the reported 1.39mm error by Shu .et.al. This is useful for registering small objects in surgery where it is not feasible to place a marker/tracker on them.
- 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.
While the methodology is useful in post-operative analysis of implant stability, potential for enhancing precision of plate registration intra-operatively remains a topic for debate. Since the “registration” of digital to real plate is done after the holes have been drilled and the crews have been placed, this method will not be suitable as a navigation aid in the critical phase of choosing the correct positioning of the plate.
- 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 mention open access to source code or data but provides a clear and detailed description of the algorithm to ensure 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
Experiment is too rigid and does not consider real life scenarios: The bone was fixed beforehand and thus factors such as the variation on the position of the bones might affect the screw and plate placement, which will affect the way the participants hold the instrument, which might lead to occlusion of the marker from the view of the Atracsys / tracking quality. During operation, placement of the bone-fiducial is dependent on the surgeon, and these placements might affect how the virtual bone is being registered onto the real bone. The experiment does not take that into account. Tracking of instruments to guide screw or pin has already been done, such as „Validation of mixed-reality surgical navigation for glenoid axis pin placement in shoulder arthroplasty using a cadaveric model.“ by Sanchez-Sotelo et.al.
- 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 — could be rejected, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
Interesting method to register virtual plate with the real plate without direct use of marker, but the clinical application based on the methodology is not wide enough outside of plate alignment assessment after the plate has been already fixed onto the bone. Live application for guidance (which is on of the main pros for digital twin use) does not seem to be aplicable with this method.
- Reviewer confidence
Very confident (4)
- [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.
I continue to see the limited clinical relevance, in my opinion, as the main weakness of this study, which was also not sufficiently resolved in the rebuttal. The author highlights the specific value for intraoperative use, i.e verification of screw position alignment and plate position, in some case it might be useful as an alternative to X-Ray/ CT based post-procedure validation. However, the usefulness of this method depends on the clinical perspective: It is more beneficial for the patient if the screws and plates guided to be placed correctly during the procedure, and follow up with some form of validation before closing up the patient if needed, or rather than hoping that the surgeon aligns the screws and plate correctly, discovered that they have misaligned the placement, and repeat the procedure again in order to realigned the screws and plate. The latter would ultimately bear the greater harm to the patient. Therefore, while there is some use in the method proposed as an alternative to X-Ray/ CT for post procedure validation, which could eliminate exposure to radiation, unless the author explicitly guides the reader to the above mentioned value in the introduction and discussion, it feels like putting the cart before the horse, and seem counter-intuitive.
Review #3
- Please describe the contribution of the paper
The paper presents a method for converting from tracking data for instruments in orthopaedic surgery to a digital twin for screw and plate implants. The main challenge here is to determine when and where placements occur for screws and plates, so that the drill track and model can provide the screw position and orientation. For this purpose, the manuscript proposes an activity recognition method based on LSTM networks, followed by deterministic algorithms to determine the placement of screws and the plates they are attached to. A phantom bone study with 28 surgeons evaluated the method, and with RMSE of 1.52 mm and 0.94 mm for screws and plates, respectively.
- 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 well-written and clearly motivated.
- The application of activity recognition to the problem of DT update for orthopaedic surgery is novel and interesting.
- In the large phantom study (28 physicians) the final accuracy for plate placement is good.
- 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 fundamental assumption is weak, that the drill pose is determinative of the screw pose. Phantom bones with no soft tissue avoid many of the challenges of real surgery, in which the drill pose may be a poor indicator of the final screw orientation.
- 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 does not mention open access to source code or data but provides a clear and detailed description of the algorithm to ensure 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
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, the paper has significant novelty, integrating two fields of research (activity recognition and digital twins) in a novel way. The biggest issue is the assumption that the drill pose is determinative of the screw pose, and in practice there is not a clear way to validate the final digital twin. Cadaveric studies are needed in future work to validate the method in a more realistic setting, where soft tissue and other factors may affect the accuracy of the method. The results are promising, and although the evaluation is limited to a phantom study, they are sufficiently strong to warrant acceptance at MICCAI.
- Reviewer confidence
Very confident (4)
- [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.
Overall, the points raised by other reviewers have not changed my opinion of the paper, and I still believe it is an accept.
With regard to R3, I actually found the comparisons to be quite comprehensive. While perhaps not exhaustive in this conference paper, they are sufficient to demonstrate the value of the proposed approach over SOTA baselines.
With regard to speed, the proposed DyMamba has inherent advantages in the linear complexity compared to transformers, which does not need to be demonstrated in explicit runtime experiments.
Author Feedback
We thank all reviewers for their constructive feedback. We are pleased that the clear motivation (R1, R4) and the novelty (R1, R4) of our approach, the phantom study (R1, R4), and its resulting accuracy (R3, R4) were appreciated. We clarify the key points below: Generalizability and Clinical Importance of Implant Integration (R1): We agree that generalizability and clinical relevance were underemphasized. Our method is adaptable to other plating scenarios (e.g. distal radius, proximal humerus, tibial shaft fractures), as procedural steps are consistent. This enables broad clinical use, including intraoperative decision support, postoperative assessment, treatment planning, or simulation-based training with automatic evaluation using objective metrics. The introduction and discussion will be revised to better emphasize these aspects. Limitations for Intraoperative Guidance (R3): We acknowledge limitations in intraoperative guidance during early plate positioning. Our method enables real-time screw reconstruction—including length and insertion depth estimation—which provides actionable feedback and extends the approach by Sanchez-Sotelo et al. (2023). However, plate reconstruction is triggered only after three screws have been placed, making navigation during initial plate placement infeasible. This constraint is inherent to our indirect tracking approach, designed to be minimally invasive and user-friendly. Nonetheless, the method offers intraoperative value, such as verifying screw selection and alignment, or confirming plate position before closure. This limitation will be addressed in the revised discussion. Assumption of Drill Pose Determining Screw Pose (R4): This was well noted—using the drill pose to approximate screw pose introduces a source of error. However, this assumption was explicitly tested and quantified in our phantom study: the RMSE between the reconstructed and the actual physical screw positioning was 1.52 mm, demonstrating strong potential. In the discussion, we will propose future work to evaluate the approach in real surgical settings. Clarification on Training Data and Skill-Level Analysis (R1): We appreciate the opportunity to clarify a source of confusion. The LSTM was trained on one hour of screwing activity performed by engineers; the 11 surgeons participated only in a separate phantom-based validation. This will be clarified more explicitly. As there was no overlap between training and validation data, we did not expect skill-related bias. Therefore, no subgroup analysis was conducted. Translation into Clinical Utility (R1): Several reviewers commented on the method’s clinical translation. The following three responses address the main aspects highlighted. Tracking Occlusions (R1, R3, R4): We agree that occlusions—due to crowded ORs (R1, R3) or soft tissue (R4)—are a key challenge, as also observed in our phantom study. Although we used the Atracsys Fusion Track 250 for evaluation, it is not part of our method, which is hardware-agnostic and supports emerging solutions (e.g. markerless tracking or sensor fusion). The discussion will outline how future work may integrate such technologies for clinical robustness. Intraoperative Bone Registration (R3): We acknowledge that intraoperative registration of bone fragments is challenging. However, this area has seen substantial progress (e.g. 2D–3D registration). Such techniques are compatible with our method and could be integrated in future work to enable evaluation under realistic conditions. Bone Deformation (R1): We agree that material differences between synthetic and human bone may affect deformation. However, Synbones are more flexible than real bones (Zdero, 2023), suggesting deformation is unlikely to be more problematic clinically. Editorial Improvements (R1): We appreciate the suggestions and will implement all changes, including the participant description, referring to all individuals as participants, with training-level composition specified.
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”.
While the paper has linitations, it presents an interesting CAI approach with potential practical applications. The reviewers have some disagreement, some of which can be addressed in the rebuttal.
- 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
Meta-review #2
- 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’
The clinical relevance/significance remains the main issue and was not addressed post-rebuttal, as pointed out by R1/R3.
Meta-review #3
- 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’
This paper proposes a method to create a digital twin of orthopedic implants using 6D tool tracking combined with machine learning to estimate screw and plate placement during fracture fixation. While the study includes a relatively large phantom evaluation with 28 participants and demonstrates promising registration accuracy, the clinical relevance and novelty felt limited to the reviewers. They also raised several key weaknesses: the experimental setup is very rigid and limited to a fixed sequence of three screws and one plate, with unclear generalizability to other implant types or real surgeries; the assumption that drill pose directly determines screw pose is questionable, especially given the lack of soft tissue and realistic surgical conditions; occlusion of tracking markers remains a significant problem, which the authors do not adequately address; no subgroup analysis was done despite varied user skill levels; and the method provides plate registration only after screws are placed, limiting its intraoperative guidance value during critical initial placement. The authors’ rebuttal clarifies some points but does not fully resolve concerns about clinical translatability, robustness in live settings, or broader applicability. Overall, I would lean toward rejection.