| Diffusion Tensor Magnetic Resonance Imaging (DT-MRI) provides promising information about the anatomical structure of human brain. One important and widely studied topic is fiber tract detection, which plays important roles in diagnosing neuro-diseases, analyzing brain anatomical structures and so on. The conventional tools include constructing diffusion ellipsoid images and introducing single voxel based measurable scalars. The former one provides visible but rough sense about the local area water diffusion properties in brain tissues. However, it is not numerically evaluable and lacks the capability of making statistical inference. Existing approaches for the latter one are either non-quantitative or single voxel based approaches and vague in detection criteria. This dissertation proposes a local test-based approach for detecting anisotropic water diffusion areas in human brain. Under mild regularity conditions, we demonstrate that the proposed test statistic asymptotically follow chi 2 distribution under null hypotheses which provides clear calibration criteria. To yield improved performance of the proposed test statistic for limited resolution experiment data, an adaptive neighborhood selection approach for selecting reasonable neighborhood for each voxel of interest is proposed and used in numerical studies. Furthermore, simulations and real data applications show that the proposed test-based approach offers higher identification strength for anisotropic areas than the popularly used scalar FA (RA). |
