Print Close

RNN-based Meta-Matching for Translating rs-fMRI-Phenotypic Prediction from Big to Small Datasets

Poster No:

1122

Submission Type:

Abstract Submission

Authors:

Chen Zheng¹, William Johnstone¹, Narun Pat¹

Institutions:

¹University of Otago, Dunedin, New Zealand

First Author:

Chen Zheng
University of Otago
Dunedin, New Zealand

Co-Author(s):

William Johnstone
University of Otago
Dunedin, New Zealand

Narun Pat
University of Otago
Dunedin, New Zealand

Introduction:

The limited number of subjects in neuroimaging datasets poses a significant challenge for applying machine learning to predict phenotypes from neuroimaging data. A recently proposed solution, called meta-matching (He et al., 2022), offers a promising technique to address this limitation. However, the current implementation of meta-matching for resting-state fMRI uses a classical Artificial Neural Network and does not treat resting-state fMRI as a time series. In this study, we benchmarked the use of Bi-directional Long Short-Term Memory (Bi-LSTM) (Schuster & Paliwal, 1997), which treats fMRI as a time series, to translate phenotype prediction of cognitive abilities from a large dataset (N=11k) to a smaller dataset (n=1k).

Methods:

This study utilized a Bi-LSTM model to extract features from the resting-state fMRI time series data. The model was pre-trained to predict several general phenotypes from the Adolescent Brain Cognitive Development (ABCD) dataset (Casey et al., 2018). Then, a transfer learning technique was applied to the pre-trained models for the downstream dataset, Human Connectome Project Young Adults (HCP-YA) (Van Essen et al., 2013), to obtain meaningful embeddings. These embeddings were used to construct test-set correlation matrices for Kernel Ridge Regression (KRR) to predict different phenotypic measures.

Results:

Our preliminary results indicate that Bi-LSTM with meta-matching demonstrates enhanced predictive performance on phenotypes in the HCP-YA dataset, leading to an out-of-sample correlation that predicts cognitive abilities at r = 0.2422.

Conclusions:

Our preliminary findings suggest that Bi-LSTM is a promising algorithm for meta-matching transfer learning in fMRI BOLD time-series predictive analysis.

Modeling and Analysis Methods:

Classification and Predictive Modeling ¹

fMRI Connectivity and Network Modeling ²

Keywords:

FUNCTIONAL MRI

Machine Learning

Other - Deep Learning, Bi-LSTM, RNN

^1|2Indicates the priority used for review

Abstract Information

By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.

I accept

The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:

I am submitting this abstract as an original work to be reproduced. I am available to be the “source party” in an upcoming team and consent to have this work listed on the OSSIG website. I agree to be contacted by OSSIG regarding the challenge and may share data used in this abstract with another team.

Please indicate below if your study was a "resting state" or "task-activation” study.

Resting state

Healthy subjects only or patients (note that patient studies may also involve healthy subjects):

Healthy subjects

Was this research conducted in the United States?

Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.

Not applicable

Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.

Not applicable

Please indicate which methods were used in your research:

Computational modeling

Provide references using APA citation style.

He, T., et al. (2022). Meta-matching as a simple framework to translate phenotypic predictive models from big to small data. Nature Neuroscience, 25(6), 795–804. https://doi.org/10.1038/s41593-022-01059-9
Schuster, M., & Paliwal, K. K. (1997). Bidirectional recurrent neural networks. IEEE Transactions on Signal Processing, 45(11), 2673–2681. https://doi.org/10.1109/78.650093
Casey, B. J., et al. (2018). The Adolescent Brain Cognitive Development (ABCD) study: Imaging acquisition across 21 sites. Developmental Cognitive Neuroscience, 32, 43–54.
Van Essen, D. C., et al. (2013). The WU-Minn Human Connectome Project: An overview. NeuroImage, 80, 62–79. https://doi.org/10.1016/j.neuroimage.2013.05.041

UNESCO Institute of Statistics and World Bank Waiver Form

I attest that I currently live, work, or study in a country on the UNESCO Institute of Statistics and World Bank List of Low and Middle Income Countries list provided.