Wang J, Xiao L, Wilson TW, Stephen JM, Calhoun VD, Wang YP. Examining brain maturation during adolescence using graph Laplacian learning based Fourier transform.
J Neurosci Methods 2020;
338:108649. [PMID:
32165231 DOI:
10.1016/j.jneumeth.2020.108649]
[Citation(s) in RCA: 3] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [What about the content of this article? (0)] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/24/2019] [Revised: 01/21/2020] [Accepted: 02/23/2020] [Indexed: 11/23/2022]
Abstract
BACKGROUND
Longitudinal neuroimaging studies have demonstrated that adolescence is a crucial developmental period of continued brain growth and change. Motivated by both achievements in graph signal processing and recent evidence that some brain areas act as hubs connecting functionally specialized systems, we propose an approach to detect these regions from a spectral analysis perspective. In particular, as the human brain undergoes substantial development throughout adolescence, we evaluate functional network difference among age groups from functional magnetic resonance imaging (fMRI) measurements.
NEW METHODS
We treated these measurements as graph signals defined on the parcellated functional brain regions and proposed a graph Laplacian learning based Fourier transform (GLFT) to transform the original graph signals into the frequency domain. Eigen-analysis was conducted afterwards to study the behaviors of the corresponding brain regions, which enabled the characterization of brain maturation.
RESULT
We first evaluated our method on the synthetic data and then applied it to resting state and task fMRI data from the Philadelphia Neurodevelopmental Cohort (PNC) dataset, comprised of normally developing adolescents from 8 to 22 years of age. The method provided an accuracy of 94.9% in distinguishing different adolescent stages and we detected 13 hubs from resting state fMRI and 16 hubs from task fMRI related to brain maturation.
COMPARISON WITH EXISTING METHODS
The proposed GLFT demonstrated its superiority over conventional graph Fourier transform and alternative graph Fourier transform with high predictive power.
CONCLUSION
The method provides a powerful approach for extracting brain connectivity patterns and identifying hub regions.
Collapse