Fiber Tracking For Brain White Matter

White matter fiber tracking based on diffusion tensor imaging data is an important method for analyzing the structure of brain and is also one important base for brain disease diagnosis and surgical planning. At present, though there are many white matter fiber tracking algorithms, white matter fiber tracking, especially the tracking of complex fiber structure, is still a challenging task in medical imaging processing, due to partial volume effects and noise problem.Based on the analysis of current research situation, this paper firstly studies and implements a deterministic fiber tracking method based on the fourth order Runge-Kutta. Then, an improved probabilistic tracking method is presented combined with high order tensor model in order to solve the problem and overcome shortcomings of non-deterministic fiber tracking method. Compared with similar methods, the improved probabilistic tracking method improves the performance of crossing fibers tracking. In addition, this paper analyzes the fiber tracking method based on unscented kalman filter and presents the UKF-HOT method that combines UKF and HOT, and takes use of the performance of high order tensor model on the framework of UKF. Experimental results on simulated data and real data show that the proposed methods in this paper have advantages in dealing with complex fiber structure and have practical value on analysis of the nerve fibers in the brain. The main work is as follows:(1)The deterministic fiber tracking method is implemented based on fourth order Runge-Kutta and the relevant experimental results are given.(2)An improved probabilistic tracking method is proposed by analyzing the Friman’s probabilistic tracking method. Based on the framework of bayesian probabilistic tracking, this method calculates posterior probability according to the orientation distribution function of high order tensor model, and improves the capabilities of processing crossing fibers.(3)The UKF-HOT method is proposed after analyzing Malcolm’s fiber tracking method based on UKF. This method combines UKF’s capabilities of target tracking and HOT’s capabilities of voxel modeling, where independent coefficients of high order tensor are used as the state of UKF system, and fibers are tracted by following the state changes of the UKF system. Compared with similar methods, the UKF-HOT method improves the accuracy of fiber tracking and shows better performance on complex fiber structures.(4)Lastly, the application of the proposed fiber tracking methods in medical image processing are analyzed and discussed with examples.