Estimating dynamic brain functional networks using multi-subject fMRI data

Identifying brain regions contained in brain functional networks and functions of brain functional networks is of great significance in understanding the complexity of the human brain. The 160 regions of interest (ROIs) in the human brain determined by the Dosenbach’s template have been divided into six functional networks with different functions. In the present paper, the complexity of the human brain is characterized by the sample entropy (SampEn) of dynamic functional connectivity (FC) which is obtained by analyzing the resting-state functional magnetic resonance imaging (fMRI) data acquired from healthy participants. The 160 ROIs are clustered into six clusters by applying the K-means clustering algorithm to the SampEn of dynamic FC as well as the static FC which is also obtained by analyzing the resting-state fMRI data. The six clusters obtained from the SampEn of dynamic FC and the static FC show very high overlap and consistency ratios with the six functional networks. Furthermore, for four of six clusters, the overlap ratios corresponding to the SampEn of dynamic FC are larger than that corresponding to the static FC, and for five of six clusters, the consistency ratios corresponding to the SampEn of dynamic FC are larger than that corresponding to the static FC. The results show that the combination of machine learning methods and the FC obtained using the blood oxygenation level-dependent (BOLD) signals can identify the functional networks of the human brain, and nonlinear dynamic characteristics of the FC are more effective than the static characteristics of the FC in identifying brain functional networks and the complexity of the human brain.

Download Full-text

Computing personalized brain functional networks from fMRI using self-supervised deep learning

10.1101/2021.09.25.461829 ◽

2021 ◽

Author(s):

Hongming Li ◽

Srinivasan Dhivya ◽

Zaixu Cui ◽

Chuanjun Zhuo ◽

Raquel E. Gur ◽

...

Keyword(s):

Deep Learning ◽

Resting State ◽

Brain Function ◽

Resting State Fmri ◽

Fmri Data ◽

Functional Networks ◽

Brain Functional Networks ◽

Human Connectome Project ◽

Functional Homogeneity ◽

Decoder Architecture

ABSTRACTA novel self-supervised deep learning (DL) method is developed for computing bias-free, personalized brain functional networks (FNs) that provide unique opportunities to better understand brain function, behavior, and disease. Specifically, convolutional neural networks with an encoder-decoder architecture are employed to compute personalized FNs from resting-state fMRI data without utilizing any external supervision by optimizing functional homogeneity of personalized FNs in a self-supervised setting. We demonstrate that a DL model trained on fMRI scans from the Human Connectome Project can identify canonical FNs and generalizes well across four different datasets. We further demonstrate that the identified personalized FNs are informative for predicting individual differences in behavior, brain development, and schizophrenia status. Taken together, self-supervised DL allows for rapid, generalizable computation of personalized FNs.

Download Full-text

Group ICA for Identifying Biomarkers in Schizophrenia: ‘Adaptive’ Networks via Spatially Constrained ICA Show More Sensitivity to Group Differences than Spatio-temporal Regression

10.1101/429837 ◽

2018 ◽

Cited By ~ 1

Author(s):

Mustafa S Salman ◽

Yuhui Du ◽

Dongdong Lin ◽

Zening Fu ◽

Eswar Damaraju ◽

...

Keyword(s):

Classification Performance ◽

Fmri Data ◽

Group Differences ◽

Support Vector ◽

Functional Networks ◽

Adaptive Networks ◽

Brain Functional Networks ◽

Subject Specific ◽

Spatio Temporal ◽

Constrained Ica

AbstractBrain functional networks identified from fMRI data can provide potential biomarkers for brain disorders. Group independent component analysis (GICA) is popular for extracting brain functional networks from multiple subjects. In GICA, different strategies exist for reconstructing subject-specific networks from the group-level networks. However, it is unknown whether these strategies have different sensitivities to group differences and abilities in distinguishing patients. Among GICA, spatio-temporal regression (STR) and spatially constrained ICA approaches such as group information guided ICA (GIG-ICA) can be used to propagate components (indicating networks) to a new subject that is not included in the original subjects. In this study, based on the same a priori network maps, we reconstructed subject-specific networks using these two methods separately from resting-state fMRI data of 151 schizophrenia patients (SZs) and 163 healthy controls (HCs). We investigated group differences in the estimated functional networks and the functional network connectivity (FNC) obtained by each method. The networks were also used as features in a cross-validated support vector machine (SVM) for classifying SZs and HCs. We selected features using different strategies to provide a comprehensive comparison between the two methods. GIG-ICA generally showed greater sensitivity in statistical analysis and better classification performance (accuracy 76.45±8.9%, sensitivity 0.74±0.11, specificity 0.79±0.11) than STR (accuracy 67.45±8.13%, sensitivity 0.65±0.11, specificity 0.71±0.11). Importantly, results were also consistent when applied to an independent dataset including 82 HCs and 82 SZs. Our work suggests that the functional networks estimated by GIG-ICA are more sensitive to group differences, and GIG-ICA is promising for identifying image-derived biomarkers of brain disease.

Download Full-text

Functional Anatomy of Intensity Aspects of Attention

Zeitschrift für Neuropsychologie ◽

10.1024/1016-264x.14.3.181 ◽

2003 ◽

Vol 14 (3) ◽

pp. 181-190 ◽

Cited By ~ 2

Author(s):

Walter Sturm

Keyword(s):

Right Hemisphere ◽

Functional Anatomy ◽

Fmri Data ◽

Functional Networks ◽

Top Down ◽

Stimulus Modality ◽

Spatial Orienting ◽

Phasic Alertness ◽

Co Activation ◽

Attention System

Abstract: Behavioral and PET/fMRI-data are presented to delineate the functional networks subserving alertness, sustained attention, and vigilance as different aspects of attention intensity. The data suggest that a mostly right-hemisphere frontal, parietal, thalamic, and brainstem network plays an important role in the regulation of attention intensity, irrespective of stimulus modality. Under conditions of phasic alertness there is less right frontal activation reflecting a diminished need for top-down regulation with phasic extrinsic stimulation. Furthermore, a high overlap between the functional networks for alerting and spatial orienting of attention is demonstrated. These findings support the hypothesis of a co-activation of the posterior attention system involved in spatial orienting by the anterior alerting network. Possible implications of these findings for the therapy of neglect are proposed.

Download Full-text