Paper Info Reviews Meta-Review Author Feedback Post-rebuttal Meta-Reviews Back to top
List of Papers By topics Author List
Paper Info Reviews Meta-review Author Feedback Post-Rebuttal Meta-reviews

Authors

Marilyn Keller, Marcell Krall, James Smith, Hans Clement, Alexander M. Kerner, Andreas Gradischar, Ute Schäfer, Michael J. Black, Annelie Weinberg, Sergi Pujades

Abstract

To treat bone fractures, implant manufacturers produce 2D anatomically contoured plates. Unfortunately, existing plates only fit a limited segment of the population and/or require manual bending during surgery. Patient-specific implants would provide major benefits such as reducing surgery time and improving treatment outcomes but they are still rare in clinical practice. In this work, we propose a patient-specific design for the long helical 2D PHILOS (Proximal Humeral Internal Locking System) plate, used to treat humerus shaft fractures. Our method automatically creates a custom plate from a CT scan of a patient’s bone. We start by designing an optimal plate on a template bone and, with an anatomy-aware registration method, we transfer this optimal design to any bone. In addition, for an arbitrary bone, our method assesses if a given plate is fit for surgery by automatically positioning it on the bone. We use this process to generate a compact set of plate shapes capable of fitting the bones within a given population. This plate set can be pre-printed in advance and readily available, removing the fabrication time between the fracture occurrence and the surgery. Extensive experiments on ex-vivo arms and 3D-printed bones show that the generated plate shapes (personalized and plate-set) faithfully match the individual bone anatomy and are suitable for clinical practice.

Link to paper

DOI: https://doi.org/10.1007/978-3-031-43990-2_46

SharedIt: https://rdcu.be/dnwL1

Link to the code repository

https://humerusplate.is.tue.mpg.de/

https://github.com/MarilynKeller/HumerusPlate

Link to the dataset(s)

N/A

Reviews

Review #1

  • Please describe the contribution of the paper

    The paper is concerned with the optimal design of humerus implants. The paper investigates the use of a statistical shape model of the humerus for modifying the shape of the humerus implant by deformation of an ‘ideal’ design on the reference geometry. The deformations are post processed to ensure anatomical regions map to the same anatomical region. The humerus design are evaluated by an expert surgeon with cadaveric and 3D printed samples and by quantitative assessment.

  • 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 clinical need is clearly defined and the proposed method extensively considers the application of this method to the clinical need.
    • The authors extensively review relevant literature
    • The authors demonstrate that their approach that uses well established methods, statistical shape modelling and registration can generate patient specific humerus geometries that would be surgically useful.
    • The implants are extensively evaluated: They are evaluated ex vivo in cadavers, with simulated geometries using 3D printing, and virtually with quantitative assessments of distances.
  • 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 main weakness of the work is a lack of a clear definition of what ‘optimal’ means for the humerus design. This should be more thoroughly defined. o The authors state that <2mm deviation in fixation areas and <5mm in other areas. Are we to understand that lower is better? o What is the target deviation of the implant from the bone surface? How is the target accuracy related to uncertainty in reconstructing the damaged humerus. o How accurately can the implants be placed and how does this factor into the analysis?
    • There is a lack of detail in many of the descriptions that make understanding challenging and reproducing the method challenging.
      o Statistical shape modelling: what dataset is used? This is important for interpreting the evaluations presented. Is the SSM evaluated in the same or different datasets.  When evaluating the bone registration errors it is unclear if the evaluations are on the images used to build the SSM. If they were used in the SSM it is surprising that the results are not more accurate. Or is it the case that these errors reflect the errors introduced by Bi vs B’i? Is there any characterization of Bi vs B’i? o Why did the authors use MD to characterize registration accuracy rather than Hausdorff distance which is more commonly used.
    • The methodological novelty is limited. But the focus of the manuscript is the fusion of many pieces and application to a clinical problem to prove out translational potential.
  • Please rate the clarity and organization of this paper

    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
    • good things about reproducibility o the paper uses established methods o the paper has extensive supplemental material
    • limiting the ability to reproduce o a closed dataset is used o many aspects of the data processing, organization and algorithm lack sufficient detail for reproducibility
  • 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 figures could all be improved.
      o Most captions do not adequately describe the figure. The reader can not understand the figures without reference to the text.
      o The colour map used for distance uses multiple colours and it is hard to interpret without having a colour bar on the figure. The ranges seem to often have 5mm as the maximum, however 5mm is the edge of acceptable. If 5mm and higher are all mapped to red it is impossible to tell if the red values are acceptable or not. Singel colours with differences in shade are preferred because of there easier interpretation. o Labelling of the specimens shown in figure 4 is required for interpretation. Can you maintain the same humerus pose for all figures to make comparison easier.
  • 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 paper is focused on optimizing plate shape but the definition of optimal is not well defined. The approach and results are well presented.

  • 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 addresses a relevant challenge in preoperative orthopedics surgical planning. Shape analysis methods are used to build a pipeline that suggest patient-specific humerus plate for a corrective osteotomy. Methods have been evaluated on several datasets including CT scans, 3D printed models as well as 3 cadaver specimens, and the obtained results show the feasibility of the method.

  • 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.
    • Clinical relevance
    • Overall good methodology based on shape analysis
    • A promising method for the design an placement of patient-specific humerus tailored plates
  • 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 shape variation between the on-the market plates and the proposed customized-shape plates is unclear. Several fixation or surgical-related constraints might hinder the clinical transfer.
    • Comparison to SOTA methods is missing
    • Lack of shape descriptors or semantic metrics which might be necessary for a better “fit criteria”
  • Please rate the clarity and organization of this paper

    Satisfactory

  • 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
    • Datasets are not available.
    • Source code is not available.
    • Implementation details are partially described.
  • 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 paper addresses a relevant challenge in preoperative orthopedics surgical planning. The supplementary materials allow a good understanding of the plate set creation and placements algorithms.

    General remark concerning the structure of the introduction section: The introduction section can be better organized by a comprehensive summary of the key related works, followed by their limitations and finally how does the present method addresses some or all the pointed limitations. The contributions (which are placed at the top of page 2) can be moved to the last paragraph in the introduction section. The “tutorial” of the planning steps can be presented at the beginning methods sections. The introduction is not the most suitable section to include details regarding the evaluation “In addition, we perform a case study evaluation with an expert surgeon, in which the validity of the numerical criterion is confirmed.”
    Further, please have a look a related work [Carrillo et al., 2017].

    Specific remarks/questions:

    1. Figures:
      • Fig.1 The blue color in (middle) is misleading because the shaft appears in blue and then in the (right), the shaft appears in (pale)red.
      • Are there 2 mean shapes (atlases) one for each of the groups A & B?
      • Fig.3 Left Do A and B correspond to those 2 mean shapes?
      • Fig.4 Right - Why there exist 7 bones instead 6?
    2. Methods: The shape analysis and all the algorithms are well explained in the paper and supplementary materials. Still, few points remain unclear to me:
    • Page 3. “From these annotations, we design an ideal plate shape contoured to a humerus template bone mesh T. This plate is designed to have similar dimensions to the actual INTEOS PROXIMAL HUMERAL PLATE 3.5.”
      -> What does ideal plate shape mean in this context. Does it has always the same dimensions of the mentioned plate? It would also be good to insert a reference for the cited plate model or brief details.
    • Equation 1. “ F is a 3D pervertex offset applied to each mesh vertex” -> How to define F?
    • Page 5. Plate extraction the optimal plate design is transferred to the registered bone using a per-vertex index basis. Does the registered bone has the same number of vertices and indices after being registered to the template bone? In case a “correspondence” step has been performed to have the number of vertices in the registered bone the same as the vertices in the atlas, how the plate transfer can be then achieved?
    1. Evaluation:
      • Few quantitative results are mentioned in the manuscript (colors are not enough). It would be good to provide details about the bones, the atlases and the plate sets such as their dimensions and the number of vertices/elements.

    • It might be a good idea to reconsider the fitting function and reformulate it including more shape descriptors and global error metrics instead of the discrete numeric values (2mm & 5mm).

    • It would be also reasonable to consider the pipeline as a decision support system with quantitative information and augmented visualization. Having a “binary decision” might not be preferable for some surgeons.

    • Except for the atlas availability, are there any limits having the method suitable for left side humerus as well?

    • Please insert the references in ascending id order within the text.

    [Carillo et al., 2019] Carrillo, F., Vlachopoulos, L., Schweizer, A., Nagy, L., Snedeker, J., Fürnstahl, P. (2017). A Time Saver: Optimization Approach for the Fully Automatic 3D Planning of Forearm Osteotomies. In: Descoteaux, M., Maier-Hein, L., Franz, A., Jannin, P., Collins, D., Duchesne, S. (eds) Medical Image Computing and Computer-Assisted Intervention − MICCAI 2017. MICCAI 2017. Lecture Notes in Computer Science(), vol 10434. Springer, Cham. https://doi.org/10.1007/978-3-319-66185-8_55

  • 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 proposed solution is good. Still, many considerations and refinement are required to show a significant benefit over the existing systems. Further, the qualitative validation is performed by a single surgeon who takes part in the planning the procedure as well, and more comparison to baseline methods and/or retrospective planning cases are encouraged.

  • 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 #4

  • Please describe the contribution of the paper

    This work proposes optimal custom plate shape generation and plate-to-bone positioning methods. It also proposes using a custom set plate (N number of plates) for a population. Furthermore, the automated position system can be used to evaluate a plate’s compatibility. Performs analysis of the proposed methods based on experiments on ex-vivo arms and 3D-printed bones.

  • 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.

    1) The automatic custom plate generation method can be used to generate patient-specific plates. 2) Proposes building a set of plates using a greedy algorithm (to have an optimal number of plates in a set) to accommodate a bone population. 3) Proposes plate-to-bone position methods that can also be used to evaluate a plate’s compatibility for surgery. 4) The generated custom plates and set plates are analysed using ex-vivo arms and 3D-printed bones.

  • 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.

    1) Limited quantitative analysis: This work lacks a p-value study to show statistical significance on any/all observations (made based on the binary decision criteria).
    2) Unclear/ Limited Novelty: a. It is unclear how the custom plate is auto-generated. Page 3, section 2, the author states that “From these annotations, we design an ideal plate shape contoured to a humerus template bone mesh T.” From here, it is inferred that an optimal plate is hand-designed for a template bone. While it is clear how a new bone is registered to the template bone, it is unclear how the new bone-to-template bone registration results in a custom plate (shape/length). What shapes and lengths are changed? based on what factor? and how? b. Limited clarity on if this work is a naïvely combination of both [13] and [10], or if there are any additional technical novelties for custom plate generation. 3) Given the lack of plate stiffness/stress analysis or observations on patient comfort, it seems too early to argue for “relaxation of the theoretical fitting constraints” in the discussion. In the absence of such studies, at least a study on the widely employed “hand-bend plate” technique should be performed to observe if they adhere to those constraints strictly, and if not, are there any noticeable discomfort/ side effects to patients. 4) All 4 figs qualities need to be improved (refer to constructive comments for the authors) 5) Writing: The flow of writing could be improved for better paper organization and improve ease of reading. a. Introduction Para 2 could be used as the last para in the introduction section. b. While employing prior techniques [13] and [10], The author could have added their description and key differences in the methodology section “prelims” instead of describing them in the introduction section. The reader needs to constantly read the second half of the introduction section and methodology to get a gist. 6) In the evaluation based on 3D printed bones, as the surgeon only chose the plate same as the plate proposed by the algorithm twice (out of 7), and it is argued that both (surgeon and algorithm chosen) plates were fit for surgery, an analysis on “fit for surgery” between the bone and all plates in the set could give a valuable insight. For instance, did all plates in the set were fit for surgery or was the plate proposed by the algorithm among the top-3 choices of the surgeon for each bone?

  • Please rate the clarity and organization of this paper

    Satisfactory

  • 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

    No codes provided in the supplimentary.

  • 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) Add p-value to each claim/observation to show statistical significance. 2) Improve figures quality: a. Fig 1: i. left: Add legends to surgeon’s annotations on the bone area (contact with plate vs area to avoid) and increase the visibility of the bone area (The bottom bone surface is too dark). ii. Middle: Highlight the offset region. iii. Use region marking (color stripes) to show regions that need <2mm distance and <55mm distance instead of area.
    b. Figure 2: i. Left: The Si color doesn’t look orange. [very minor]. ii. Right: Annotate or provide legends for different colors. What do they represent? c. Figure 3: i. Caption: “Set A (top) and set B(bottom)” seems incorrect. It should be set A (left) and set B (right). d. Figure 4: i. Left: It is unclear which ones are custom plates and which ones are set plates. ii. Right: The centre part of the plates cannot be seen. Please follow Fig 3 right. 3) Please edit the introduction and methodology to improve the flow of information. Also, clearly state how a custom plate is generated based on bone-bone template registration. 4) Report the operating time to show any significant advantage between using a custom/set plate template vs a hand-contoured plate. This could highlight the advantage in terms of surgery time.

  • 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

    4

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

    This work proposes novel ideas and performs an ex-vivo and 3D-printed bone study. However, this study lacks any clear quantitative analysis (with statistical significance). The qualitative analysis could also be interpreted in both ways (eg: the surgeon not choosing the same plate as the algorithm could also be interpreted as there is no clear advantage of the proposed method) and leaves the reader with many questions. Taking into consideration these factors, and the additional work needed to improve both writing and figure quality, I recommend weak reject.

  • Reviewer confidence

    Somewhat confident

  • [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed

    5

  • [Post rebuttal] Please justify your decision

    The clarification on technical novetly, clarification on surgeon’s choice of plate against the SOTA plates and clinical need (author has declared that they will make parts of this work public for non-commercial use) was significant enough to change by rating.



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.

    The paper proposes to develop a custom humerus plate which matches the 3D shape of a bone for minimally invasive surgery. It’s an interesting work. However, there are some limitations. Compared with other methods in Introduction and Experiment should be added. The organization of the paper can be improved.



Author Feedback

We thank R1, R3, and R4 for their constructive feedback. We are happy that they appreciated our novel approach to creating personalized plates with a clear clinical application, as well as the ex-vivo and 3D printed evaluations. We next clarify some items.

  • Plate optimality The optimality of the plate design (R1) is defined by a surgeon (p3 Plate design), establishing fixation points, areas to avoid, tolerances and validated by a second surgeon (p4 l.2). A plate fulfilling these constraints on a bone is considered optimal for surgery. We will make this “optimality definition” explicit in the Plate design paragraph.

  • Statistical shape model R1, R3: on the humerus statistical shape model (SSM): We build the SSM on the bones from group A only (R1). The mean bone of A is the only atlas. With the SSM we register all bones (Groups A and B). Figure 2 left shows: (top) a noisy scan from group B in pink, and the registered bone in green superimposed; (bottom) the registered bone alone. We will clarify this in the caption. The scans (in pink) have holes and extra pieces of bone attached, making Hausdorff distances not representative of the registration quality. In Fig. 3 left we report per vertex max distances, which highlights these noisy areas in the B dataset (caption will be fixed).

  • Plate design R3, R4: on the plate extraction process: The biggest confusion is raised by R4-2a on how the optimal plate is extracted from the humerus. Our plate geometry is defined from the humerus template surface. The plate-to-bone offset F is then recorded. When we register this template to a new bone, we can then generate a plate with the same offset F, that covers the same region of the plate (p5 “plate extraction”). We will clarify p5 with this offset definition in the camera-ready version and add the plate dimensions and the number of vertices of the atlas.

  • Related work Regarding related work and comparisons, R1 writes “authors extensively review relevant literature”. The related work discussion is summarized in Tab. 1 of Sup. Mat. R3: we do compare the fit of our plate to the SOTA plate in Fig. 3. R4: About the contribution compared to [13] and [10]. The registration method of [10] does not preserve anatomic regions, and smooth shells [4] does not work on raw scans (p4 last sentence). [13] extracts a plate design from a mean bone mesh. In contrast, we extract the plate from any registered mesh, generalizing to any bone shape. Moreover, we propose a plate set approach that was not considered in [13]. We will add the Carrillo reference, thanks R3.

  • R4 final claim We would like to emphasize that R4’s final statement “the surgeon not choosing the same plate as the algorithm could also be interpreted as there is no clear advantage of the proposed method” is misleading. In p8, the surgeon had the choice among several plates, which are all designed with the proposed method. As written in p8 3rd paragraph the surgeon picked a different one among the new plate set, as “different plates with slightly different placements can work for the same bone”. We did not specify it, but the surgeon always preferred to use the plates of our set rather than the SOTA plates, as also shown by the experiment in Fig. 3 right. We thank R4 and will clarify this in p8.

  • Reproducibility The reviewers pointed out missing implementation details. We will release for non-commercial use i) the humerus SSM, ii) the source code for the plate extraction, and iii) the code for optimizing the position of a plate on a bone. This will provide implementation details (R3) and allow reproducibility

  • Figures Constructive feedback on how to improve figures (colors, bone pose, color-coded bar, captions) will be easily integrated into the final version, e.g. (R3) the seven bones in Fig 4 are explained in the first sentence of 3.3, we will fix the caption.



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.

    We appreciate the authors’ novel approach to creating personalized plates with a clear clinical application. The research is interesting. The authors provide some satisfactory explanations in response to the reviewers’ comments. If the authors make portions of their work publicly available, it could greatly benefit clinical practice.



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 paper presents a novel approach for optimizing the 3D plate shape for proximal humerus fractures, employing statistical shape modeling to design patient-specific humerus implants. The reviews exhibit general consensus on the clinical relevance and innovation of the paper. The main discrepancies arise from the perceived weaknesses and the extent to which the authors’ responses alleviate these concerns. The rebuttal clarified many methodological and implementation details; however, it did not fully address all the issues raised, specifically around the paper’s novelty and statistical analysis. The concerns about the statistical analysis and the clear distinction from previous works ([13] and [10]) were not entirely addressed.



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.

    Several issues were pointed out by the reviewer. I suggest a revision and resubmission at another venue.



back to top