Variability of Non-parametric HRF in Interconnectedness and Its Association in Deriving Resting State Network
Document Type
Conference Proceeding
Publication Date
1-1-2023
Abstract
Blood Oxygen Level-Dependent (BOLD) time course in functional magnetic resonance imaging (fMRI) is modeled as the response of the hemodynamic response function (HRF) excited by an activity-inducing signal. Variability of the HRF across the brain influences functional connectivity (FC) estimates and some approaches have been attempted to separate the HRF and activity-inducing signal from the observed BOLD signal as a blind separation problem. In this work, an approach based on homomorphic filtering is proposed to estimate a non-parametric representation of HRF in resting state fMRI. Voxel-wise and region-wise variations of correlation of the estimated HRF (both the parametric and non-parametric representation) are analyzed in different functional networks. Principal component analysis of the correlation matrix using the estimated HRF is used to analyze the interconnectedness. HRF shows higher variability for the non-parametric representation over the parametric representation. Further, the contribution of the estimated HRF is then studied in producing resting-state networks using the dictionary learning framework.
Identifier
85172416379 (Scopus)
ISBN
[9783031430749]
Publication Title
Lecture Notes in Computer Science Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics
External Full Text Location
https://doi.org/10.1007/978-3-031-43075-6_21
e-ISSN
16113349
ISSN
03029743
First Page
239
Last Page
248
Volume
13974 LNAI
Grant
2021B1515020019
Fund Ref
National Science Foundation
Recommended Citation
Das, Sukesh Kumar; Jain, Pratik; Sao, Anil K.; and Biswal, Bharat, "Variability of Non-parametric HRF in Interconnectedness and Its Association in Deriving Resting State Network" (2023). Faculty Publications. 2148.
https://digitalcommons.njit.edu/fac_pubs/2148