| Summary: | 碩士 === 國立臺灣大學 === 醫學工程學研究所 === 90 === The information of size and orientation of organismic cells is very important for the study of microstructures and detection of the related diseases. Diffusion MRI is the only technique that can non-invasively provide such information. For example, diffusion tensor magnetic resonance imaging (DTI) is capable of providing the fiber tract information of the white matter in the brain. However, due to the interference of the noise, identifying fiber tract in a DTI remains as an unresolved problem in general. In this thesis, we propose a new algorithm, which may successfully determine the fiber tract in a DTI and overcome the potential mis-guidance caused by the noise. The proposed algorithm is composed of four essential techniques, namely, fractional anisotropy, Hough transform, depth first search, and dynamic programming. The fiber tract determined by the proposed algorithm has been shown to be highly consistent with that in the Mn2+ enhanced T1WI.
The q-space diffusion weighted MRI, derived by using multiple b-values, has been developed to obtain the structural information for material science and biological system. In this thesis, we have devised a new phantom to validate the accuracy of the microstructure sizes attained by using the signal decay curve in q-space imaging. The experimental results show that we are able to derive the microstructure sizes accurately. Besides, from the diffusion probability density function of the water molecules in q-space imaging, we are able to obtain several new types of contrast images based on such physical parameters of the water diffusion as the mean displacement, the probability at zero displacement and the self-diffusion coefficient of the water molecules. This technique has been applied to the biological systems and phantoms, respectively, in this study.
In the attempt to combine the advantages of the DTI and q-space imaging, a novel technique, called q-space diffusion tensor imaging (q-DTI), has been proposed in this study, which can map size and orientation of the microstructures at the same time. The effect of the SNR and b-value on imaging parameters has been investigated by using phantoms. Owing to its well-known neural structures, the rat Corpus Callosum has been used as a ‘test pattern’ to verify the q-DTI in defining intersecting fibers. The phantom and rat models can be further employed to optimize q-DTI imaging scheme. The limitations and the potential applications of the q-DTI technique have been discussed.
|
|---|
