Intrinsic dynamic effective connectivity analysis of the resting-state network in COPD
Poster No:
1562
Submission Type:
Abstract Submission
Authors:
Jinho Bae1, Kyung-II Han2, Tae Hyung Kim2, Hang Joon Jo3
Institutions:
1Hanyang Univ., Seoul, Korea, Republic of, 2College of Medicine, Hanyang, Seoul, Korea, Republic of, 3Hanyang University, Seoul, AK
First Author:
Co-Author(s):
Introduction:
A common chronic respiratory disease known as chronic obstructive pulmonary disease (COPD) is characterized by persistent respiratory symptoms and irreversible airflow limitation, and cognitive impairment is a frequent and essential comorbidity in COPD patients. Although functional connectivity (FC) can be beneficial for describing abnormal patterns of brain activity, it cannot be used to infer the underlying effective connectivity (EC), which is defined as the directional causal relationships among brain regions. The purpose of this study was to explore the dynamic changes in the intrinsic effective connectivity of the resting-state (RS) network and their relationship with cognitive impairment in COPD patients.
Methods:
At the time of the study, 50 COPD patients over the age of 60 without a precise diagnosis of cognitive dysfunction or depression were asked to take part. Healthy controls with no respiratory symptoms who were matched for sex and age with the patients were recruited. Of the 50 COPD patients participating in the study, 38 were chosen as the patient group. Twelve patients who were excluded had severe brain shrinkage or infarction. Thus, this study included 38 patients with COPD and 30 healthy controls. For all subjects, anatomy images were acquired by a T1-weighted MRRAGE pulse sequence. Then, RS fMRI data were acquired with a gradient echo-planar pulse sequence, and they were preprocessed by an automatic pipeline of AFNI (afni_proc.py). Networks of 14 functional regions of interest (ROIs) were used as a mask from the atlas of 90 functional ROIs created by Stanford Greicius Lab. We conducted three types of smoothing processes: (i) without smoothing, (ii) gaussian smoothing, and (iii) smoothing within the ROI mask, and we extracted the mean time series, 95% principal component (PC), and first PC time series for each functional ROI network. Vector auto-regression (VAR) is a multivariate forecasting algorithm that can be used when two or more time series influence each other. We performed VAR (1) modeling, which contained up to one lag of each predicted time series, with 90 observed time series influencing each other as input. Through this model, a coefficient matrix was estimated to determine how the past value of each time series affects the current value of the other time series. To identify the directionality, strength, and lag effect of EC within each group (normal and patient groups), a 1-sample t-test was performed on the coefficient matrix estimated in the VAR (1) model. Through this, it was evaluated whether the relationship between time series within each region was statistically significant. In addition, a 2-sample t-test was performed on the coefficient matrix of each group to evaluate the difference in time series between the normal group and the patient group. Through this, it was evaluated whether there was a statistically significant difference in the dynamic interaction of time series between the two groups. In addition, the correlation between networks with differences between groups, and cognitive assessment was analyzed in COPD patients.
Results:
Compared to normal groups, there were significant differences among COPD patients between and within networks (fig. 1). In addition, there was a significant correlation with some essential clinical indicators (FEV1%_ac) in COPD patients. The significance threshold was set at Bonferroni-corrected p < 0.05 (fig. 2).
Conclusions:
We attempted to identify dynamic changes in the intrinsic EC of the RS network. For this purpose, the VAR model was used, and the analysis was conducted on the premise that, due to the nature of the VAR model, mutual influences must exist between all input time series data. It is necessary to check whether all-time series data affect each other and whether their connectivity is statistically significant. In addition, the normality of the time series data should be verified to have the same number of observations.
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 1
fMRI Connectivity and Network Modeling 2
Keywords:
Other - chronic obstructive pulmonary disease ; effective connectivity ; resting-state fMRI ; vector auto-regression
1|2Indicates the priority used for review
Provide references using author date format
Gang Chen. (2011), 'Vector autoregression, structural equation modeling, and their synthesis in neuroimaging data analysis', Computers in Biology and Medicine, vol. 41, no. 12, pp. 1142-1155