Authors

Xinrui Yuan, Jiale Cheng, Fenqiang Zhao, Zhengwang Wu, Li Wang, Weili Lin, Yu Zhang, Gang Li

Abstract

During the early postnatal period, the human brain undergoes rapid and dynamic development. Over the past decades, there has been increased attention in studying the cognitive and cortical development of infants. However, accurate prediction of the infant cognitive and cortical development at an individual-level is a significant challenge, due to the huge complexities in highly irregular and incomplete longitudinal data that is commonly seen in current studies. Besides, joint prediction of cognitive scores and cortical morphology is barely investigated, despite some studies revealing the tight relationship between cognitive ability and cortical morphology and suggesting their potential mutual benefits. To tackle this challenge, we develop a flexible multi-task framework for joint prediction of cognitive scores and cortical morphological maps, namely, disentangled intensive triplet spherical adversarial autoencoder (DITSAA). First, we extract the mixed representative latent vector through a triplet spherical ResNet and further disentangles latent vector into identity-related and age-related features with an attention-based module. The identity recognition and age estimation tasks are introduced as supervision for a reliable disentanglement of the two components. Then we formulate the individualized cortical profile at a specific age by combining disentangled identity-related information and corresponding age-related information. Finally, an adversarial learning strategy is integrated to achieve a vivid and realistic prediction of cortical morphology, while a cognitive module is employed to predict cognitive scores. Extensive experiments are conducted on a public dataset, and the results affirm our method’s ability to predict cognitive scores and cortical morphology jointly and flexibly using incomplete longitudinal data.

Link to paper

DOI: https://doi.org/10.1007/978-3-031-43996-4_52

SharedIt: https://rdcu.be/dnwPU

Link to the code repository

N/A

Link to the dataset(s)

N/A

Reviews

Review #3

Please describe the contribution of the paper
1. A novel multi-task framework for joint prediction of cortical morphological maps and cognitive development scores has been proposed.
2. An attention-based feature disentanglement module has been proposed.
3. The application is clinical significant.
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 encoder is with attention-based disentanglement. The identity conditional block is also integrated into the network to preserve identity-level regression/progression pattern. Cognition prediction module and cortical morphology prediction module are all designed with prior knowledge. The proposed method has been validated on a publicly available dataset, which would be an asset of the research community. The experimental results are convincing.
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 authors may consider to compare more state-of-the-art methods. Because the three methods reported in Table 2 are not directly applied to the same applications (cognitive and cortical morphological preperty prediction tasks) as the proposed method did. Since the proposed method only tested on one dataset, whether the proposed method can be used for other dataset or application is unknown.
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

The authors did not provide the source code. But the authors clearly described the building blocks and loss function used in the paper.
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

This paper presents a novel auto-encoder-based deep learning network. The network is composed novel building blocks. The domain knowledge has been incorporate into the network design. The authors also proposed novel loss function. The paper is easy to follow. The paper is well-organized. In the reviewer’s opinion, the paper still needs some improvements. The authors need to strengthen the experimental results by comparing more recent state-of-the-art methods. The authors may want to assess the proposed method on more than one dataset. In addition, the authors may want to share the code for reproducing by other researchers.
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

6
Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

The technical novelties is great. The application is of great clinical significance. The prior knowledge from the clinical application is incorporated into the method design which is a good way to distinguish MICCAI papers from CVPR or ICCV papers.
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 #1

Please describe the contribution of the paper

The authors present a multi-task framework that predicts cognitive score, cortical morphology change, and age for infants based on cortical surface at a given time point, in infant brains
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 paper addresses an important issue of regularization of longitudinal infant brain scan data. Proposed methods may benefit both clinical patients and research community.
2. The paper is clearly structured.
3. The approach of the paper is systematic and the problem at hand is addressed from multiple viewpoints. It represents an impressive body of work.
4. Comparison to previous methods is performed.
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. Evaluation is performed on a single dataset.
2. No detail is provided on implementation and performance of the proposed framework, which raises questions of both reproducibility and usability of this development.
3. Figures are not self-explanatory, contain a lot of fine print and are in general hard to parse.
Please rate the clarity and organization of this paper

Very 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

The authors go at lengths to provide verbal descriptions of the used underlying developments, and explaining the organization of their framework. No source code is provided. Thus, the reproducibility can be estimated as slightly above average.
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

Overall this is a good piece of work, described in good detail. My primary concerns are: 1) Please make figures (especially Fig. 1) more readable by providing verbal descriptions of symbols used, and explaining color/shape coding. 2) Detail on underlying implementation/source code or Docker image would help the reader to estimate the feasibility/practicability of your approach to their specific application.
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

7
Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

The strengths of the paper significantly outweigh its weaknesses. Compared to the average, this is a strong body of important work presented in good detail.
Reviewer confidence

Very confident
[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

The paper proposes a multi-task framework for joint prediction of cognitive scores and cortical morphological maps of infants. Towards this end, the contributions are three-fold: (1) an attention-based feature disentanglement module to separate the identity- and age-related features from mixed latent features in order to effectively extract the discriminative information at individual-level and form the basis for dealing with irregular and incomplete longitudinal data; 2) a novel identity con-ditional block to fuse identity-related information with designated age-related infor-mation, which can model the regression/progression process of brain development flexibly; 3) a unified, multi-task framework to jointly predict the cognitive ability and cortical morphological development and enable flexible prediction at any time points during infancy by concatenating the subject-specific identity information and identity conditional block.
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.
- Can handle longitudinal data scanned irregularly within 24 months of age. This is important because longitudinal infant images are usually collected at diverse and irregular scan ages in practice.
- Approach designed using a combination of SOTA AI along with biological knowledge
- Achieves SOTA performance
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 model is not tested on any external dataset for generizability assessment.
- The language is ok but there are rooms of improvement.
Please rate the clarity and organization of this paper

Very 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

Tested on a public dataset, and the authors have mentioned in the reproducibility checklist that the codes will be made available.
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 authors are advised to test their approach on an independent dataset to assess the generalizability of the model
- The authors mention “statistically significant” improvement in results in several places but have not mentioned which statistical test(s) was used to do these comparisons.
- It is interesting that the proposed approach performs slightly worse than “w/o AFD” for predicting FMS (Table 1). Any thoughts on that can be included in the discussion.
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?

As stated before
Reviewer confidence

Somewhat confident
[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

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 is well structured, with innovative methods and targets important missing time-point issue in longitudinal studies. More details such as the figure descriptions could be provided.

Author Feedback

N/A

back to top

Multi-task Joint Prediction of Infant Cortical Morphological and Cognitive Development