List of Papers Browse by Subject Areas Author List
Abstract
Acute allograft rejection poses a significant challenge in kidney transplantation, the primary remedy for end-stage renal disease. Timely detection is crucial for intervention and graft preservation. A notable obstacle involves ensuring consistency across Diffusion Weighted Magnetic Resonance Imaging (DW-MRI) scanning protocols at various Tesla levels. To tackle this, we propose a novel, non-invasive framework for automated diagnosis of acute renal rejection using DW-MRI. Our method comprises several key steps: Initially, we register the segmented kidney across different scanners, aligning them from the cortex to the medulla. Afterwards, the Apparent Diffusion Coefficient (ADC) is estimated for the segmented kidney. Then, the ADC maps are partitioned into a 3D iso-surface from the cortex to the medulla using the fast-marching level sets method. Next, the Cumulative Distribution Function (CDF) of the ADC for each iso-surface is computed, and Spearman correlation is applied to these CDFs. Finally, we introduce a Transformer-based Correlations to Classes Converter (T3C) model to leverage these correlations for distinguishing between normal and acutely rejected transplants. Evaluation on a cohort of 94 subjects (40 with acute renal rejection and 54 control subjects) yields promising results, with a mean accuracy of 98.723%, a mean sensitivity of 97%, and a mean specificity of 100%, employing a leave-one-subject testing approach. These findings underscore the effectiveness and robustness of our proposed framework.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2024/paper/3658_paper.pdf
SharedIt Link: pending
SpringerLink (DOI): pending
Supplementary Material: N/A
Link to the Code Repository
N/A
Link to the Dataset(s)
N/A
BibTex
@InProceedings{Abd_ANew_MICCAI2024,
author = { Abdelhalim, Ibrahim and Abou El-Ghar, Mohamed and Dwyer, Amy and Ouseph, Rosemary and Contractor, Sohail and El-Baz, Ayman},
title = { { A New Non-Invasive AI-Based Diagnostic System for Automated Diagnosis of Acute Renal Rejection in Kidney Transplantation: Analysis of ADC Maps Extracted from Matched 3D Iso-Regions of the Transplanted Kidney } },
booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2024},
year = {2024},
publisher = {Springer Nature Switzerland},
volume = {LNCS 15012},
month = {October},
page = {pending}
}
Reviews
Review #1
- Please describe the contribution of the paper
This paper proposes a novel, non-invasive diagnostic framework for automated diagnosis of acute renal rejection in kidney transplantation using diffusion-weighted magnetic resonance imaging (DW-MRI). The system extracts ADC maps from matched 3D iso-regions of the transplanted kidney and computes the cumulative distribution function (CDF) of ADC for each iso-surface. The system then leverages the correlations of CDFs to distinguish between normal and acutely rejected transplants. The proposed framework is validated on a cohort of 94 subjects and achieves a mean accuracy of 98.723%, a mean sensitivity of 97%, and a mean specificity of 100%. The system demonstrates the potential to improve the accuracy and efficiency of acute renal rejection diagnosis in kidney transplantation.
- 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.
-
Strong results
-
Good methodology, this is an interesting and useful application of diffusion-weighted MR imaging and should be discussed at the conference
-
- 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.
-
While the results on the small dataset show very promising results, when translating any technology to clinical level, a critical factor is the failure assessment of the methodology, there is not discussion in the paper about the failure or how it can be detected. Also the 94 subjects is quite a small dataset. Perhaps methods like uncertainty quantification could be helpful here to detect when the DL based approach is not giving a confident result
-
While leave out one subject is a valid method, a test set analysis would certainly help in building more confidence for the method
-
The method relies on good kidney segmentation, while there are solutions available, its quite possible that the entire methodology could fail if the kidney is not segmented well enough
-
- 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.
- Do you have any additional comments regarding the paper’s reproducibility?
No
- 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
Refer weaknesses
- 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 Accept — could be accepted, dependent on rebuttal (4)
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
I think this work has a strong application based story that deserves a discussion at the conference. It is also a unique case of diffusion-weighted MR application and few works as such exist in the literature so far
- Reviewer confidence
Very confident (4)
- [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 paper introduces an innovative AI-based diagnostic system utilizing a Transformer-based model, called T3C, to analyze 3D iso-regions of ADC maps from kidney transplants. It is a non-invasive framework to diagnose acute renal rejection, enhancing the precision of early detection.
- 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 use of Transformer-based models to correlate features within ADC maps is novel in the context of renal transplantation diagnostics. The reported accuracy, sensitivity, and specificity are exceptionally high, suggesting that the model performs well in the dataset. Therefore, this system may provide a tool for early and accurate detection of acute renal rejection which may improve patient outcomes significantly by allowing earlier interventions.
- 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 dataset seems to be collected by the author themselves, but lacks more detailed descriptions. For example, how to divide this dataset in the experiment? how much data does training, validation, and testing each contain?
In the experiment, T3C far exceeded the comparison methods by nearly double. But these comparison methods are all traditional machine learning methods. To my knowledge, training transformer models typically requires a relatively large dataset. In the experiment, there are only 97 patients. How many correlation matrices are generated after processing these data for the transformer to learn? In addition, the specific parameters of the transformer are not provided in the experiment, such as the number of heads, the size of the embedding, etc. Therefore, while the paper introduces a novel approach, more detailed exploration and discussion on choosing transformers instead of other network models will strengthen the research findings.
- 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 details about dataset, for example, how to divide this dataset in the experiment? how much data does training, validation, and testing each contain? Provide more details about parameters, including the T3C and other machine learning methods. This may increase the credibility of the experimental results. Provide more discussion about the results. Explain why the transformer is able to achieve nearly twice the performance of traditional machine learning methods in this task. Besides, the paper may benefit from grammar modifications, such as unifying tenses.
- 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 Accept — could be accepted, dependent on rebuttal (4)
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The non-invasive framework for automated diagnosis of acute renal rejection is innovative. However, the paper lacks some details about dataset and the models. Also, while the transformer is able to achieve nearly twice the performance of traditional machine learning methods in this task, the discussions are missing.
- Reviewer confidence
Very confident (4)
- [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
This paper proposed to classify acute kidney rejection from MRI ADC maps, computed from 11 b values with b0 as the base. A cohort from multiple-devises of 94 patients were processed. Traditional ML algorithms and transformer-based algorithms were used to have achieved promising results and performed correlation analysis. Larger cohorts and multi-modalities are provided as the future work directions for better accuracy and more insights.
- 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.
Strengths of the paper are about the novelty in data scale in the emerging area of kidney transplants, methodology and experimental design are reasonable and into details for reproducibility, from registration, to ML model parameters, to transformer-based PCA correlation analysis.
- 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.
Weaknesses of the paper are more insignts and possibly directions are needed in this emerging area.
- Please rate the clarity and organization of this paper
Excellent
- 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?
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
This paper proposed to classify acute kidney rejection from MRI ADC maps, computed from 11 b values with b0 as the base. A cohort from multiple-devises of 94 patients were processed. Traditional ML algorithms and transformer-based algorithms were used to have achieved promising results and performed correlation analysis. Larger cohorts and multi-modalities are provided as the future work directions for better accuracy and more insights.
Strengths of the paper are about the novelty in data scale in the emerging area of kidney transplants, methodology and experimental design are reasonable and into details for reproducibility, from registration, to ML model parameters, to transformer-based PCA correlation analysis.
Weaknesses of the paper are more insignts and possibly directions are needed in this emerging area.
- 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 studied an emerging area of kidney transplants and has collected a data cohort of 94 patients. The propsed methodology is reasonable for the calssification task and correlation analysis. Future work of larger cohort and utilization of multi-modalities are also promising.
- 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.