List of Papers Browse by Subject Areas Author List
Abstract
This work investigates the use of configuration state imaging together with deep neural networks to develop quantitative MRI techniques for deployment in an interventional setting. A physics modeling technique for inhomogeneous fields and heterogeneous tissues is presented and used to evaluate the theoretical capability of neural networks to estimate parameter maps from configuration state signal data. All tested normalization strategies achieved similar performance in estimating T2 and T2*. Varying network architecture and data normalization had substantial impacts on estimated flip angle and T1, highlighting their importance in developing neural networks to solve these inverse problems. The developed signal modeling technique provides an environment that will enable the development and evaluation of physics-informed machine learning techniques for MR parameter mapping and facilitate the development of quantitative MRI techniques to inform clinical decisions during MR-guided treatments.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2024/paper/0605_paper.pdf
SharedIt Link: https://rdcu.be/dV58O
SpringerLink (DOI): https://doi.org/10.1007/978-3-031-72120-5_47
Supplementary Material: N/A
Link to the Code Repository
https://github.com/fuslab-uofu/mri-signal-model
Link to the Dataset(s)
N/A
BibTex
@InProceedings{Ada_Physics_MICCAI2024,
author = { Adams-Tew, Samuel I. and Odéen, Henrik and Parker, Dennis L. and Cheng, Cheng-Chieh and Madore, Bruno and Payne, Allison and Joshi, Sarang},
title = { { Physics informed neural networks for estimation of tissue properties from multi-echo configuration state MRI } },
booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2024},
year = {2024},
publisher = {Springer Nature Switzerland},
volume = {LNCS 15011},
month = {October},
page = {502 -- 511}
}
Reviews
Review #1
- Please describe the contribution of the paper
This paper uses a physics informed, fully connected neural network to estimate MR relaxivity and flip angle from multi-echo configuration state MRI. The authors demonstrated that the choice of normalization approach significantly impacts the estimation performance.
- 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 author did an excellent job describing the underlying physical model from the MR physics standpoint. The network architecture and training strategies are also well described.
- 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.
Despite the novelty of the MR sequence, the idea of physics informed neural network is not new. The network architecture is also a simple MLP.
- 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
Please see my comments in the strengths and weaknesses sections. For in-vivo data, the authors may consider a convolutional network architecture to incorporate the spatial information for better fitting.
- 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 novelty of this work’s deep learning method approach is limited. The authors applied the well-known concept of physics informed neural network for parameter estimation using the novel multi-echo configuration state MRI. I am not sure whether it is suitable for this conference.
- 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
A method is proposed to estimate multiple tissue MRI parameters (T1, T2, T2*, FA map) from a dedicated sequence (unbalanced steady state) proving mixed information encoded in so-called configuration states. A physics-informed neural network is proposed for the parameter estimation from the measured configuration state signals. Training is done on simulated data, testing on simulated and experimental 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.
The main contribution is the application of PINNs to solve a complex signal equation for the acquired MRI data, which would otherwise be untractable. The proof of concept is shown on simulated and ex-vivo data, with comparison to ground truth T1 and T2 maps.
- 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 results are still preliminary, so it is difficult to assess the robustness of the method in another scenario, where the assumptions on the range of parameter values may be different, where other factors (not present in the signal model) may have a significant contribution (e.g. motion/flow, magnetization transfer etc.), or simply when the noise is higher.
- 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 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
Could the authors clarify how the input data (the configuration states F_{-2}, F_{-1} etc…) are derived from the acquired data (is it different echoes?)? Otherwise, the reference is necessary. Note: reference [1] shouldn’t have be anonymized, instead the authors should have cited it in a passive/impersonal form like: “we used a sequence such as that described in [1]”. It was not very clear to me why the normalization that was found to be optimal in the simulation (pathway) was not used in the actual experiment (sample), could the authors comment? Could the authors add details about the acquisition sequence and parameters for the actual experiment, as well as the acquisition time?
- 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 proposed acquisition technique is an original way to acquired multiparametric MRI data. The novel contribution is to use PINNs to estimate the parameters, whereas a classic inversion of the signal model would be very hard. Some missing information could be added to strengthen the paper, as suggested.
- 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 manuscript shows that using a simulated set of values a network can be trained to mostly correctly estimate T1, T2, T2* and FA parameters. This shows that a neural network is able to dies
- 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 presented concept is original in it’s approach even tough it does resemble the concept of MRI-Fingerprinting. The description of methods is a particularly strong element of the work, which makes it clear how the experiments were conducted and what were the approaches in this work.
- 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 paper is the lack of verification of the method in either an in vivo experiment (where the resulting values could be compared against previously published results) or an in vitro experiment using more standardized reference phantoms (for example agarose doped with NiCl2).
- 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 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?
It would be beneficial if the authors shared the simulation code in the supplement or in open access however it is understandable if they refrain from it.
- 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
A discussion of the salt-pork phantom results are clearly missing in the work. Were the parameters achieved during imaging in the typical range for this type of phantom? It could be also interesting to discuss other influential factors in the signal which the simulated approach didn’t take into account (such as magnetization transfer described by Bloch-McConnel equations). Lastly the method should be discussed in reference to MRI-Fingerprinting in regards to advantages and disadvantages
- 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?
Overall this is a very sound paper, with a very good execution. Few open questions are left in regards to the discussion but no major issue point toward the necessity to reject the work
- 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
Author Feedback
We would like to thank the reviewers for their thoughtful evaluation of the manuscript. We will incorporate small requested clarifications into the final submission. Additionally, the signal modeling code will be made open source and available on GitHub.
Meta-Review
Meta-review not available, early accepted paper.