Abstract

Magnetic Resonance Imaging (MRI) is widely used in diagnosing anterior cruciate ligament (ACL) injuries due to its ability to provide detailed image data. However, existing deep learning approaches often overlook additional factors beyond the image itself. In this study, we aim to bridge this gap by exploring the relationship between ACL rupture and the bone morphology of the femur and tibia. Leveraging extensive clinical experience, we acknowledge the significance of this morphological data, which is not readily observed manually. To effectively incorporate this vital information, we introduce ACLNet, a novel model that combines the convolutional representation of MRI images with the transformer representation of bone morphological point clouds. This integration significantly enhances ACL injury predictions by leveraging both imaging and geometric data. Our methodology demonstrated an enhancement in diagnostic precision on the in-house dataset compared to image-only methods, elevating the accuracy from 87.59% to 92.57%. This strategy of utilizing implicitly relevant information to enhance performance holds promise for a variety of medical-related tasks.

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2024/paper/3065_paper.pdf

SharedIt Link: pending

SpringerLink (DOI): pending

Supplementary Material: https://papers.miccai.org/miccai-2024/supp/3065_supp.pdf

Link to the Code Repository

N/A

Link to the Dataset(s)

N/A

BibTex

@InProceedings{Liu_ACLNet_MICCAI2024,
        author = { Liu, Chao and Yu, Xueqing and Wang, Dingyu and Jiang, Tingting},
        title = { { ACLNet: A Deep Learning Model for ACL Rupture Classification Combined with Bone Morphology } },
        booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2024},
        year = {2024},
        publisher = {Springer Nature Switzerland},
        volume = {LNCS 15005},
        month = {October},
        page = {pending}
}


Reviews

Review #1

  • Please describe the contribution of the paper

    The study proposes a model that integrates convolutional MRI image representations with transformer-based bone morphological point clouds to predict ACL rupture by leveraging both imaging and geometric 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.

    Combination of imaging features with bone morphological point clouds, leveraging both imaging and geometric data.

  • 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 study lacks comparisons with existing literature. Additional ablation studies involving alternative deep learning models like ResNet and varying numbers of blocks in DenseNet would strengthen the analysis. The proposed algorithm’s reproducibility is questionable due to missing model details, including the number of blocks in DenseNet, the layer from which features were obtained, and training specifics (such as the DL model’s output and automatic segmentation model referenced as [25]). Crucial data details, such as demographics (age, gender, ACL status) and the number of subjects in each class, are missing. The absence of statistical analysis for comparison also weakens the paper’s findings. The organization of the paper could be improved to enhance readability.

  • 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.

  • Do you have any additional comments regarding the paper’s reproducibility?

    To enhance reproducibility and transparency, sharing the code used for the study and providing detailed information about the study cohort (including demographics and exclusion criteria) would be highly beneficial. Additionally, including more details about the training process would be valuable.

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review. Pay specific attention to the different assessment criteria for the different paper categories (MIC, CAI, Clinical Translation of Methodology, Health Equity): https://conferences.miccai.org/2024/en/REVIEWER-GUIDELINES.html
    • Articles involving human subjects require a statement that the studies were conducted with the approval of an Institutional Review Board (IRB) or analogous Ethics Board. However, the approval details were not provided in the paper.
    • The paper does not include a comparison of the proposed algorithm with existing literature (DOI:https://doi.org/10.1016/j.arthro.2023.08.010, doi: 10.2196/37508, https://doi.org/10.1038/s41598-024-51666-8). The performance of the existing literature appears to be better than that of the proposed model. To make a fair comparison, the authors are encouraged to apply the algorithms from existing studies to their own dataset.
    • References should be updated using the MICCAI reference template.
    • In the sentence ‘This integration significantly enhances ACL injury predictions by leveraging both imaging and geometric data’ in the abstract, the authors claim that the integration significantly improves prediction, but they did not provide the results of statistical analyses.
    • The status of ACL tear needs to be provided, since the partial tears are more difficult to classify than the complete tears.
    • The exclusion criteria need to be provided, and clarification is required regarding how the authors selected the 1904 MRI images from the total pool of 54,008 MR images.
    • In the Dataset section, the number of subjects is listed as N=1902, yet in Table 1, the total number of datasets is indicated as 1904 (1521+383).
    • It would be beneficial to include MR acquisition details in a Supplementary document, such as scanner, TE, TR, slice thickness, magnetic field strength, etc.
    • There is no need to abbreviate ACL again in the Dataset section, as it has already been abbreviated earlier.
    • In the Dataset section, the sentence ‘Sagittal MRI sequences are utilized for image data, as they provide comprehensive organizational and structural details of the anterior cruciate ligament (ACL), enhancing the model’s accuracy in characterizing ACL texture.’ suggests that the authors used sagittal MRI sequences to enhance model’s performance. However, it is unclear whether they conducted an ablation study with other imaging modalities.
    • Could the authors provide the number of medical experts who annotated the images in Dataset section?
    • It would be helpful for clarity if the authors could share the cropping method whether it was center or random cropping.
    • Could the authors provide insights into how they utilized the clinical imaging report?
    • It would be beneficial if the authors could provide the number of features obtained from the DL model.
    • Incorporating a discussion of the results would strengthen the paper.
    • Why are the results of MRNet+PCT in Table 2 and Table 4 different?
    • There are typing mistakes in the paper. It would be better to check and correct them.
    • The paper requires improvement in writing quality. The English should be thoroughly checked and corrected.
  • 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

    Weak Reject — could be rejected, dependent on rebuttal (3)

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

    The paper lacks reproducibility as data is not public, there are missing data details, DenseNet model specifications, and information on the automatic segmentation model used. Additionally, it lacks comparison with existing literature.

  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [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

    The major contribution of the paper is an innovative approach for diagnosing anterior cruciate ligament (ACL) injuries by incorporating bone morphology data into the model training process. This integration of bone morphological point clouds with MRI image data significantly enhances ACL injury predictions and improves diagnostic accuracy compared to image-only methods. Another major contribution is the dataset, which comprises 54,008 MRI images.

  • 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 novel integration of bone morphological point cloud data with MRI image data for ACL injury diagnosis.

    This paper introduces a feature fusion module that synergistically combines image features and point cloud features from two separate branches allowing the model to capture both the structural and textural essence of the ACL region from MRI images and mirror three-dimensional morphologies from point cloud data.

  • 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.

    All of the algorithm is only evluated on one dataset, which created by the authors. This paper could benefit from comparing algorithm performance on multiple dataset. Beside, in the literature review part, please include more relevant review of the related dataset and articulate the motivation of the creation of your own dataset.

  • Please rate the clarity and organization of this paper

    Very 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.

  • Do you have any additional comments regarding the paper’s reproducibility?

    N/A

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review. Pay specific attention to the different assessment criteria for the different paper categories (MIC, CAI, Clinical Translation of Methodology, Health Equity): https://conferences.miccai.org/2024/en/REVIEWER-GUIDELINES.html

    Provide more literatuer review of the related research to stress your innovation and contributation, both for the algorithm and dataset. Disscuss the limitation of this research.

  • 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

    Accept — should be accepted, independent of rebuttal (5)

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

    This paper is well written. The innovation of the algorithm and the dataset are qualify as an accepted paper. Although there are some improvement could be made, this paper is great one.

  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [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 authors demonstrate a classification method to predict anterior cruciate ligament (ACL) injuries on knee MR images. The novelty of the approach is that they include point cloud information of the bone structure obtained by manual delineation as second input. The inclusion of the extra information improves the model performance by approx. 5 %, compared to state of the art applications.

  • 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.

    -A novel method is presented including shape information of the bone. Features from MRI and the point cloud are combined using fusion methods. -The clinical application is clear, the work benefits from close clinical input. Comparison of results with manual readers. -Results are better than the compared approaches and comparable with human expert readers.

  • 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 results and contributions are not described clearly: The “Ours” method is compared with the “MRNet + PCT” (where the PCT is considered the same as in “Ours” and therefore also a contribution) -The approach incorporates more data than MRI alone, the volumetric expert annotations. Thus, an improvement in performance metrics was to be expected. Compared to this, the fully automated approach without expert input (Table 4, Ours*) shows comparable results and should therefore be the main highlighted result. -For this fully automated approach, results for the automated segmentation of cloud points are missing. -In table 4, it is unclear why “MRNet + PCT” benefits much more from the manual delineation compared to “Ours”. This is unexpected and should be justified in the Discussion & Conclusion.

  • 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.

  • Do you have any additional comments regarding the paper’s reproducibility?

    N/A

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review. Pay specific attention to the different assessment criteria for the different paper categories (MIC, CAI, Clinical Translation of Methodology, Health Equity): https://conferences.miccai.org/2024/en/REVIEWER-GUIDELINES.html

    Absolute result values would be better in the abstract, also the % is misleading (to what base number?). Besides, please note the explanations above (weaknesses). The overal structure could be improved if the fully automated approach was presented as final goal and all other as intermediate experiments. As it is now, it seems as this final experiment was not planned in the beginning, also because information is missing (segmentation results; Fig. 3 is nice but lacks an interpretation, what differences are to be seen).

  • 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

    Accept — should be accepted, independent of rebuttal (5)

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

    The novel approach and clinical argumentation together with the good results overweigh the weaknesses in structure and lead to a rather positive evaluation.

  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [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




Author Feedback

N/A




Meta-Review

Meta-review not available, early accepted paper.



back to top