Constructing the Spatial Template of Default Network

• Main Authors: Sue-Jin Lin, 林素堇
• Other Authors: Tzu-Chen Yeh
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/30670650093414457334

碩士 === 國立陽明大學 === 腦科學研究所 === 96 === There are still a lot of neuronal networks in resting human brain. One of these networks is the default network. Many studies demonstrated the existence of this network during resting state and task conditions. The network extensions included medial prefrontal cortice, posterior cingulate cortice, bilateral inferior parietal lobes, bilateral medial temporal lobes, and parahippocampal regions. The function of this brain network was still not clear. But investigators consider that default network supports the basic brain function and maintains the baseline by definitions of psychology or physiology. They also found there were some difference in functional connectivity between normal subjects, brain disease patients, and psychiatry disorder patients. The default network may provide some information to improve clinical diagnosis. Then, how to detect this brain network becomes an important issue. In the current study, we constructed two spatial templates of default network to sort the components derived by independent component analysis (ICA) algorithm which used to decompose the functional magnetic resonance imaging (fMRI) data. Protocol 1 template had limited preprocessing included slice timing and realignment before ICA. After define the default network component by meta-analysis results in raw data space and finished general linear model (GLM) estimation, the residual preprocessing have been done included co-registration, normalization, and smooth. Protocol 2 template had fully preprocessing before ICA included slice timing, realignment, co-registration, normalization, and smooth. In protocol 2, the components were selected based on meta-analysis results in normalized space. Finally, GLM estimation was applied. The group analysis showed the spatial extensions of two templates were coherent with previous published studies. But the protocol 2 template had larger spatial extensions in temporal lobes. The protocol 1 and protocol 2 templates were applied to sort the components for retest. The performances were very similar and spatial correlations were higher than meta-analysis. It seemed our templates take the advantage for the default network component selection.