Analysis of Longitudinal Change Patterns in Developing Brain Using Functional and Structural Magnetic Resonance Imaging via Multimodal Fusion

Rekha Saha, Debbrata K. Saha, Zening Fu, Marlena Duda, Rogers F. Silva, Vince D. CalhounbioRxiv 2024 · DOI

TL;DR

This study introduces novel multimodal fusion techniques to analyze how brain structure and function change together in adolescents over time. By combining fMRI and sMRI data from a large cohort, they reveal intricate patterns of brain development, highlighting sex-specific differences and links to cognitive and psychological changes.

What Problem Does This Paper Address?

Understanding how the adolescent brain develops is crucial, but individual imaging techniques like fMRI (function) and sMRI (structure) only offer partial views. These modalities can miss crucial connections and insights when analyzed in isolation. This paper tackles the challenge of integrating these complementary data sources to gain a more comprehensive understanding of longitudinal brain changes, specifically focusing on how functional and structural alterations co-occur and evolve during adolescence.

Methods

The researchers employed two primary multimodal fusion approaches:

  1. Asymmetric Fusion: This method first uses Independent Component Analysis (ICA) on functional data (fMRI) to identify "Functional Change Patterns" (FCPs). Then, it correlates these FCPs with structural data (sMRI, specifically gray matter volume - GMV) at a voxel-wise level to find structural patterns linked to functional changes.
  2. Symmetric Fusion (mCCA+jICA): This approach uses a combination of Multiset Canonical Correlation Analysis (mCCA) and Joint Independent Component Analysis (jICA). This technique simultaneously analyzes both functional (fMRI) and structural (sMRI) data to identify "Functional Change Patterns" (FCPs) and "Structural Change Patterns" (SCPs) that covary over time. This method aims to capture joint information across modalities more effectively.

Both approaches were applied to analyze changes in functional network connectivity (FNC) and gray matter volume (GMV) over a two-year period.

Datasets

The study utilized data from the Adolescent Brain and Cognitive Development (ABCD) dataset. This is a large-scale, longitudinal study of brain development in children and adolescents in the United States. The analysis specifically used data from 2,734 subjects with both baseline and two-year follow-up scans for both fMRI and sMRI.

Key Insights and Ideas

  • Brain development is a coupled process: Functional and structural changes in the brain are not independent but rather co-occur and influence each other throughout adolescence.
  • Sex matters: Females show stronger coupling between brain function and structure changes compared to males, suggesting distinct developmental trajectories.
  • Specific brain networks are key: Certain FCPs and SCPs are strongly associated with longitudinal changes in cognitive and psychopathology scores, pointing to specific neural mechanisms underlying these developmental shifts.
  • Multimodal fusion unlocks deeper insights: Combining fMRI and sMRI data through advanced techniques like mCCA+jICA reveals patterns that would be missed by analyzing each modality alone.

Key Results

  • Asymmetric Fusion: Identified FCP components 2 and 4 showing significant associations with voxel-wise GMV data.
  • Symmetric Fusion (mCCA+jICA):
    • Revealed structured FCPs, including increased connectivity in visual-sensorimotor domains and decreased connectivity in sensorimotor-cognitive control domains, linked to structural changes in the bilateral sensorimotor cortex.
  • Sex Differences: Females exhibited stronger coupling between structural change patterns (SCP component 2) and functional change patterns (FCP components 1, 3, and 4) compared to males.
  • Psychopathology & Cognition: * FCP components 4 and 5 showed a negative association with GMV changes in individuals with low psychiatric scores, suggesting stronger coupling in this group.
  • FCP components 2 and 5 showed positive and negative correlations, respectively, with GMV changes in individuals with lower cognitive scores.

Limitations and Future Work

  • Temporal Information Loss: The mCCA+jICA approach, by working with ICA components, might lose some fine-grained temporal information from the original fMRI data.
  • Linearity Assumption: The ICA method assumes linear relationships in brain functional patterns, which might not capture all complex nonlinear dynamics.
  • Age Range: The study focused on adolescents aged 9-10 years at baseline, and while psychopathology might increase later, the initial levels were relatively low.

Future work could involve:

  • Incorporating additional imaging techniques to further enhance understanding of brain changes.
  • Investigating how varying the number of components affects the identified couplings.
  • Exploring nonlinear methodologies to capture more complex brain dynamics.
  • Examining how the relationship between brain and behavior evolves in later adolescence using future waves of the ABCD data.

Summary updated

Abstract

Abstract Functional and structural magnetic resonance imaging (fMRI and sMRI) are complementary approaches that can be used to study longitudinal brain changes in adolescents. Each individual modality offers distinct insights into the brain. Each individual modality may overlook crucial aspects of brain analysis. By combining them, we can uncover hidden brain connections and gain a more comprehensive understanding. In previous work, we identified multivariate patterns of change in whole-brain function during adolescence. In this work, we focus on linking functional change patterns (FCPs) to brain structure. We introduce two approaches and applied them to data from the Adolescent Brain and Cognitive Development (ABCD) dataset. First, we evaluate voxelwise sMRI-FCP coupling to identify structural patterns linked to our previously identified FCPs. Our approach revealed multiple interesting patterns in functional network connectivity (FNC) and gray matter volume (GMV) data that were linked to subject level variation. FCP components 2 and 4 exhibit extensive associations between their loadings and voxel-wise GMV data. Secondly, we leveraged a symmetric multimodal fusion technique called multiset canonical correlation analysis (mCCA) + joint independent component analysis (jICA). Using this approach, we identify structured FCPs such as one showing increased connectivity between visual and sensorimotor domains and decreased connectivity between sensorimotor and cognitive control domains, linked to structural change patterns (SCPs) including alterations in the bilateral sensorimotor cortex. Interestingly, females exhibit stronger coupling between brain functional and structural changes than males, highlighting sex-related differences. The combined results from both asymmetric and symmetric multimodal fusion methods underscore the intricate sex-specific nuances in neural dynamics. By utilizing two complementary multimodal approaches, our study enhances our understanding of the dynamic nature of brain connectivity and structure during the adolescent period, shedding light on the nuanced processes underlying adolescent brain development.
BibTeX
@misc{saha2024analysis,
  author = {Rekha Saha and Debbrata K. Saha and Zening Fu and Marlena Duda and Rogers F. Silva and Vince D. Calhoun},
  title = {{Analysis of Longitudinal Change Patterns in Developing Brain Using Functional and Structural Magnetic Resonance Imaging via Multimodal Fusion}},
  year = {2024},
  doi = {10.1101/2024.04.07.588473},
  howpublished = {bioRxiv},
}
Figure