Research On Reconstruction Of Complex Neural Fibers

This project was supported by the National Natural Science Foundation to investigate the reconstruction of neural fibers with crossing, splaying or twisting. Some common methods as diffusion tensor imaging (DTI), Q-Ball imaging (QBI) and fourth order diffusion tensor imaging (DT4) were discussed, including basis theory, algorithm, advantages and shortcomings. The achieved results were as below:Firstly, the diffusion tensor matrix of multiple diffusion-weighted gradient directions was obtained by the multiple linear regressions. The condition number of encoding gradient matrix and standard deviation of fractional anisotropy were advanced to evaluate the image quality. Three schemes with 6, 12 and 20 gradients were used to show the influence of gradient number.Secondly, QBI was proposed to calculate the oritational distribution function. The theory of Funk-Radon transforms, spherical Gaussian interpolation and the reconstruction process using MATLAB were studied. The influence of experimental parameters as FOV and b value to SNR, angular resolution and sampling time was also investigated. Additionally, the structures of optic radiation and prefrontal cortex were listed to show the result of QBI.Thirdly, for DT4, the theory, algorithm and the expressions of fiber error and generalized anisotropy were studied. Synthetic data was introduced to investigate the effect of noise and sampling gradient number. Moreover, DTI, QBI and DT4 were compared from algorithm, experiment and imaging aspect. And the result showed that DT4 is more potential in practice.The visualizations of DTI, QBI and DT4 were finished in this project. And the reconstruction image behaves the same as the anatomical structure of human brain, which demonstrated the validity of the mentioned methods. This research could be valuable in the neural fiber tracking and clinical use.