Using Individualized Brain Network for Analyzing Structural Covariance of the Cerebral Cortex in Alzheimer's Patients

Abstract

Cortical thinning patterns in Alzheimer's disease (AD) have been widely reported through conventional regional analysis. In addition, the coordinated variance of cortical thickness in different brain regions has been investigated both at the individual and group network levels. In this study, we aim to investigate network architectural characteristics of a structural covariance network (SCN) in AD, and further to show that the structural covariance connectivity becomes disorganized across the brain regions in AD, while the normal control (NC) subjects maintain more clustered and consistent coordination in cortical atrophy variations. We generated SCNs directly from T1-weighted MR images of individual patients using surface-based cortical thickness data, with structural connectivity defined as similarity in cortical thickness within different brain regions. Individual SCNs were constructed using morphometric data from the Samsung Medical Center (SMC) dataset. The structural covariance connectivity showed higher clustering than randomly generated networks, as well as similar minimum path lengths, indicating that the SCNs are “small world.” There were significant difference between NC and AD group in characteristic path lengths (z = −2.97, p < 0.01) and small-worldness values (z = 4.05, p < 0.01). Clustering coefficients in AD was smaller than that of NC but there was no significant difference (z = 1.81, not significant). We further observed that the AD patients had significantly disrupted structural connectivity. We also show that the coordinated variance of cortical thickness is distributed more randomly from one region to other regions in AD patients when compared to NC subjects. Our proposed SCN may provide surface-based measures for understanding interaction between two brain regions with co-atrophy of the cerebral cortex due to normal aging or AD. We applied our method to the AD Neuroimaging Initiative (ADNI) data to show consistency in results with the SMC dataset.

Construction and Analysis of Weighted Brain Networks from SICE for the Study of Alzheimer's Disease

“…Random networks are computed having a random topology but sharing the size, density and binary degree distribution of the original network (Maslov and Sneppen, 2002). Cluster coefficient (Kim et al, 2016; Zhang et al, 2016) is a measure of segregation and is the average value of the cluster coefficients of each node C i computed as follows (Rubinov and Sporns, 2010): $\begin{array}{c}left{C}_i=\frac{t_i}{k_{i}false(k_i-1false)}\end{array}$ where k i is the degree of the node, and t i is the number of triangles around the node. Since the ( i, j ) element in the n 'th power of an adjacency matrix counts the number of path of length n starting at i and ending at j , the number of triangles of node i coincides with the value of the i -th diagonal element of the 3rd power of the adjacency matrix.…”

Clinically-Feasible Brain Morphometric Similarity Network Construction Approaches with Restricted Magnetic Resonance Imaging Acquisitions and their Relationship with Cognition

“…The potential role of disruption to the SCN to understanding functional outcomes has been explored within a graph theoretic framework in relation to a range of conditions. These include broad psychiatric diagnoses such as bulimia, depression and schizophrenia (Chen et al, 2017;Mak, Colloby, Thomas, & O'Brien, 2016;Palaniyappan, Park, Balain, Dangi, & Liddle, 2015;Tijms et al, 2015;Westwater, Seidlitz, Diederen, Fischer, & Thompson, 2017), neurodegenerative disorders, such as Alzheimer's disease (AD) and multiple sclerosis (Kim et al, 2016;Pereira et al, 2015;Pereira et al, 2016;Raamana, Weiner, Wang, Beg, & Alzheimer's Disease Neuroimaging, 2015;Tewarie et al, 2014), epilepsies (Garcia-Ramos et al, 2017;Sone et al, 2016;Yasuda et al, 2015) and autism spectrum disorders (Balardin et al, 2015). In all of these studies, the methodology requires multiple participants to sample enough cortical measurements to generate a correlation between all possible regional pairs.…”

“…Existing methodological approaches have attempted to investigate these structural relationships between regions at the individual-patient level (i.e. (Kim et al, 2016;Kong et al, 2015;Kong et al, 2014;Tijms, Series, Willshaw, & Lawrie, 2012;Yu et al, 2018)). The majority of these methodologies have two major limitations; they either divide ROIs into sub regions that do not respect the underlying structure and convolutions of the cortex (Tijms et al, 2012), or the edge weights are defined as the simple subtraction of the feature in region A minus region B, rather than covariance.…”

Does size matter? The relationship between predictive power of single-subject morphometric networks to spatial scale and edge weight

“…Network-level analysis of cortical thickness and gray matter features demonstrated its potential to provide novel insights or improve predictive power (Raamana et al 2015), and is gaining in popularity (Evans 2013;Wen, He, and Sachdev 2011;Reid and Evans 2013;Jason P. Lerch et al 2006). Thickness network features offer complementary information compared to the underlying fiber density (Gong et al 2012), are shown to be disrupted in AD (Kim et al 2016) and have been shown to have potential for early prognosis of AD (Raamana et al 2015;Wee et al 2012;Dai et al 2012;Kim et al 2016), as well as for subtype discrimination (Raamana, Wen, et al 2014), outperforming the non-network raw-thickness features.…”

“…Network analysis studies in cortical thickness range from 1. group-wise studies building networks based on group-wise covariance/correlation in cortical thickness (Evans 2013;He and Chen 2007;Jason P. Lerch et al 2006), which may be used to characterize the properties of these networks (such as small-worldness) as well as provide useful insight into network-level changes between two diagnostic groups e.g. healthy controls (CN) and Alzheimer's disease (AD), 2. studies building individual subject-wise graphs based on within-subject ROI-wise (pairwise) similarity metrics (Raamana et al 2015;Tijms et al 2012;Wee et al 2012;Dai et al 2012;Kim et al 2016) to enable predictive modeling. These studies resulted in disease-related insights into network-level imaging biomarkers and improved accuracy for the early prognosis of AD.…”