| Summary: | 碩士 === 國立陽明大學 === 醫學工程研究所 === 102 === Human brain is a complex neural architecture with information integration and segregation. It can be modeled as a complex network by mapping the connectivity between distinct regions. Diffusion tensor imaging (DTI) is a non-invasive technique that can probe the microstructural properties of white matter (WM) orientation with diffusion of water molecules in vivo. Streamline-tractogaphy can be utilized to reveal the WM tracts of the human brain, and further construct the WM networks. In the recent decade, human brain has been demonstrated as a “small-world” topology, i.e., high local clustering and short path lengths linking the nodes, with graph theoretic analysis, implying the optimal organization of human brain between integration and segregation. However, the WM reconstruction is highly dependent on the parameters of fiber tracking to vary the architecture of WM networks. The goal of this study is to evaluate the effect of tractography parameters for reproducibility of WM network construction.
In this study, 10 healthy subjects with no history of neurologic disease underwent two scans of DTI and T1-weighted images within one week at a 3T MR system. Fiber tracking was carried out via fiber assignment by continuous tracking with the stop criteria of a range of fractional anisotropy (FA) thresholds from 0.1 to 0.3 with an increasing interval of 0.05, and a range of curvature from 20 to 80 degree with an increasing interval of 10. Nodes of network were defined with AAL template to parcellate the gray matter (GM) into 90 regions, and the connections between regions were defined with reconstructed WM tracts. A total of 35 binary WM networks were constructed and then followed graph theoretical analysis. The intra-class correlation coefficient (ICC) was utilized to quantify the reproducibility of network metrics across the tracking parameters. Furthermore, we also constructed the WM network with a finer AAL constrained atlas (638 nodes) to estimate the properties of network at a higher resolution.
The results showed that all of binary networks were small-world networks with these selected parameters of tractogaphy. The characteristic path length (1.713) was approximately equivalent to a comparable random network, whereas the characteristic clustering coefficient (0.575) was 3.2 times greater. The topological properties showed that the constructed networks tend to random with the loose criteria. Moreover, the networks showed high reproducibility with the FA threshold between 0.2 and 0.25, and angle threshold in range of 60 to 80 degrees. A similar result was exhibited at the high resolution network. We suggest the terminated criteria with FA and angle threshold between the range with higher reproducibility and stability might be a good strategy for WM network construction with DTI-based tractography.
In this thesis, the impact of tracking parameters on WM network construction was estimated via reproducibility index, which was aimed to construct more stable, more reliable WM networks for higher repeatability and value of studies, indicating the importance of the selection of tracking parameters. This study provides more robust settings for the further WM network studies using DTI-based tractography.
|
|---|
