Estimate And Reconstruct Complex Fiber Structure Of Brain Neuro In Diffusion Magnetic Resonance Imaging

The human brain is the most complex network system in the world. The brain neuro fiber structure and the fiber connectivity between various regions were able to reflect both the variety functions and lesion information of brain. Study of human brain fiber tractography method helps to understand the micro-structure of the human brain and has important value in human brain research and brain-related diseases research. Diffusion tensor imaging(DTI) was the first technique provided an exquisite tool to study the fiber architecture of biological tissues noninvasively and in vivo, and has been largely used in medical research. This technique, despite its simplicity and robustness, has been shown to be incorrect in regions containing intravoxel orientational heterogeneity such as crossing and branching of fiber bundles. A number of methods have been proposed to overcome the single fiber orientation limitation of DTI. However, most of these methods require relatively high b-values and a large number of gradient directions to produce good results. Such requirements are generally hard to meet in common clinical research due to scan time and hardware constraints. There have been several methods, with which it is possible to resolve crossing fibers from low angular resolution diffusion imaging. However, these methods are highly sensitive to noise.This thesis around the reconstruction of complex fiber architecture and focus on the methods of intravoxel fiber architecture estimation and multifiber pathways reconstruction to improving the accuracy of complex intravoxel fiber architecture estimation and multifiber pathways reconstruction at low angular resolution diffusion imaging. The intravoxel fiber architecture estimation is the basis of multifiber pathways reconstruction. Its accuracy directly affects the credibility of the fiber path reconstructions. In low angular resolution diffusion imaging, the feature of low signal-to-noise ratio of diffusion-weighted images is more prominently exhibited, how to enhance the noise immunity of the intravoxel fiber archtecutre estimation method and improve the accuracy of the estimation is the essential of the reconstruction of complex fiber architecture. Multifiber pathways reconstruction is a prerequisite for extraction and description of the fiber connectivity, how to overcome the effects of unreliable estimation of intravoxel fiber architecture and accurately reconstruct multifiber pathways is the difficulty of the complex fiber architectures reconstruction. This thesis conducts research on the methods of intravoxel fiber architecture estimation and multifiber pathways reconstruction to overcome the limitations of the exist methods.The main research works of this thesis are as follows:(1) For the problem of noise sensing in low angular diffusion imaging, an intravoxel fiber architecture estimation method based on multitensor smoothing is proposed. The method firstly perform the initial estimation of intravoxel fiber architecture, then construct a diffusion-weighted(DW) image smoothing weighting scheme according to the properties of the multitensor field and smooth the DW images, finally reestimate the intravoxel fiber architecture to achieve an smoothed multitensor fields. The results on synthetic, physical phantom and real brain DW images show that the proposed method is able to better resolve fiber architectures while correctly preserving image edge information, thus improve the accuracy of intravoxel fiber architecture estimation.(2) For the problem of simple regularizer is difficult to fully express the prior information of fiber architecture, a nonconvex regularized blind compressed sensing method is proposed. The method is based on the multitensor model and models the DW signals as a sparse linear combination of unfixed reconstruction basis functions and introduces a nonconvex regularizer to enhance the noise immunity. We present a general solving framework to simultaneously estimate the sparse coefficients and the reconstruction basis. Experiments on synthetic, phantom, and real human brain DW images demonstrate its insensitivity to noise, its lower MAEs and its insensitivity to initialization of reconstruction basis.(3) For the three typical problems(premature termination, divagation, and inflection) of the traditional streamline tracking, an bundle constrained streamline method is proposed to accurately reconstruct multifiber pathways. Firstly, the method limits the forward direction within a reasonable range to avoid excessive deflection lead to the premature termination. Secondly, considering the neighboring fiber segment orientation, when determine the next forward direction, to avoid large deflection lead to the divagation. Finally, a fiber smoothing procedure is introduced to eliminate the inflection. Results on synthetic, physical phantom and real human brain DW images demonstrate the proposed method is able to improve the accuracy of the multifiber pathways reconstruction.