Brain functional architecture and anatomical structure have been intensively studied to generate efficient models of its complex mechanisms. Functional alterations and cognitive impairments are the most investigated aspects in the recent clinical research as distinctive traits of neurodegeneration. Although specific behaviours are clearly associated to neurodegeneration, information flow breakdown within the brain functional network, responsible to deeply affect cognitive skills, remains not completely understood. Behavioural variant Frontotemporal Dementia (bvFTD) is the most common type of Frontotemporal degeneration, marked by behavioural disturbances, social instabilities and impairment of executive functions. Mathematical modelling offers effective tools to inspect deviations from physiological cognitive functions and connectivity alterations. As a popular recent methodology, graph theoretical approaches applied to imaging data expanded our knowledge of neurodegenerative disorders, although the need for unbiased metrics is still an open issue. In this thesis, we propose an integrated analysis of functional features among brain areas in bvFTD patients, to assess global connectivity and topological network alterations respect to the healthy condition, using a minimum spanning tree (MST) based-model to resting state functional MRI (rs-fMRI) data. Contrary to several graph theoretical approaches, dependent to arbitrary criteria (e.g., correlation thresholds, network density or a priori distribution), MST represents an unambiguous modelling solution, ensuring full reproducibility and robustness in different conditions. Our MSTs were obtained from wavelet correlation matrices derived from mean time series intensities, extracted from 116 regions of interest (ROIs) of 41 bvFTD patients and 39 healthy controls (HC), which underwent rs-fMRI. The resulting graphs were tested for global connectivity and topological differences between the two groups, by applying a Wilcoxon rank sum test with a significance level at 0.05 (nonparametric median difference estimates with 95% confidence interval). The same test was applied for methodological comparison between MST and other common graph theory methods. After methodological comparisons, our MST model achieved the best bvFTD/HC separation performances, without a priori assumptions. Direct MST comparison between bvFTD and healty controls revealed key brain functional architecture differences. Diseased subjects showed a linear-shape network configuration tendency, with high distance between nodes, low centrality parameter values, and a low exchange information capacity (i.e., low network integration) in MST parameters. Moreover, edge-level and node-level features (i.e., superhighways, and node degree and betweenness centrality) indicated a more complex scenario, showing some of the key bvFTD dysfunctions observed in large scale resting-state functional networks (default-mode (DMN), salience (SN), and executive (EN) networks), suggesting an underlying involvement of the limbic system in the observed functional deterioration. Functional isolation has been observed as a generalized process affecting the entire bvFTD network, showing brain macro-regions isolation, with homogeneous functional distribution of brain areas, longer distances between hubs, and longer within-lobe superhighways. Conversely, the HC network showed marked functional integration, where superhighways serve as shortcuts to connect areas from different brain macro-regions. The combination of this theoretical model with rs-fMRI data constitutes an effective method to generate a clear picture of the functional divergence between bvFTD and HCs, providing possible insights on the effects of frontotemporal neurodegeneration and compensatory mechanisms underlying characteristic bvFTD cognitive, social, and executive impairments.
|Titolo:||Graph theory applied to neuroimaging data reveals key functional connectivity alterations in brain of behavioral variant Frontotemporal Dementia subjects|
|Data di pubblicazione:||14-feb-2020|
|Appare nelle tipologie:||8.01 Tesi di dottorato|