Human Brain Functional Aging Study Based On A Data-driven Approach

Aging is usually accompanied by progressive degeneration of all organs of the body,especially the brain,which affects memory,learning,and other cognitive functions.Although many studies have explored the mechanisms of brain aging in depth from various aspects,however,our knowledge of brain aging is still insufficient to deal with the functional degeneration or nonpathological diseases that result from brain aging.Many studies have shown that aging has an important impact on specific functional networks of the brain and also leads to a decline in the strength of functional connectivity in the brain.However,there is currently no work exploring the effects of aging on the whole brain by studying brain functional networks and the interconnections between them.The first work in this dissertation presents novel functional network and functional connectivity fusion analysis methods based on data from a large sample of 6300 healthy adults aged between 49 and 73 years from the UK Biobank project,explores their systematic covariation relationships in aging,and finds reliable aging related brain imaging markers.In this work,whole brain functional networks and its corresponding time series were first extracted based on a data-driven independent component analysis(ICA)method.And the functional network connectivity(FNC)was further calculated according to the time series of the pairwise functional networks.Next,new statistical analysis strategies are proposed to extract reliable aging related changes in functional networks and connectivity.Finally,a fusion approach is proposed to explore synergistic and contradictory changes between functional networks and connections.Results showed that the enhanced FNCs mainly occur between different functional domains,involving the default mode and cognitive control networks,while the reduced FNCs come from not only between different domains but also within the same domain,primarily relating to the visual network,cognitive control network,and cerebellum.Aging also greatly affects the functional network activation,and the increased activations along with aging are mainly within the sensorimotor network,while the decreased activations significantly involve the default mode network.More importantly,many significant joint changes between functional networks and FNCs involve default mode and sub-cortical networks.Furthermore,most synergistic changes are present between the FNCs with reduced amplitude and their linked functional networks,and most paradoxical changes are present in the FNCs with enhanced amplitude and their linked functional networks.In summary,our study emphasizes the diversity of brain aging and provides new evidence via novel exploratory perspectives for non-pathological aging of the whole brain.When extracting functional networks,we found that functional networks derived from ICA estimation tend to have much spatial noise.This is due to that although ICA’s components can be somewhat sparse due to their non-gaussian nature,the spatial sparsity and smoothness of the components are not explicitly expressed in given optimization function.To overcome this drawback in order to extract more precise functional networks and corresponding time courses from f MRI data,the second work of this paper proposes group information guided smoothing independent component analysis(GIG-SICA)by local regularization of the data to enhance the smoothness of the extracted functional networks.In order to test the effectiveness of the GIG-SICA method,our study uses Sim TB to add noise to the functional network as simulation data and input it to the GIGSICA method to estimate the functional network and its corresponding time courses.By comparing the source functional network with noise and the estimated functional network,it is proved that the GIG-SICA method can achieve the effectiveness of noise removal in most functional networks.This study also designed a comparative experiment with the group information guided independent component analysis to verify the effectiveness of the proposed new method in functional similarity and spatial smoothness.The experimental results show that the GIG-SICA method has improved to a certain extent.In summary,the GIG-SICA method can effectively extract smooth functional networks with high accuracy.On this basis,using GIG-SICA to identify brain functional networks and functional connections can provide reliable biological characteristics as a basis for further aging exploration of brain function.