Diffusion Simulation Based Tracking And Optimized Gradient Encoding Schemes For Diffusion Magnetic Resonance Imaging

Diffusion tensor magnetic resonance imaging is the only non-invasive imaging technique that allows the discrimination of different matter types as well as the distinction of matter anatomy. Anatomy investigations are possible because the measured diffusion is stronger in parallel with boundaries, such as for example cell membranes. For the reconstruction of complicated anatomies, especially the pathways of neuronal brain fibers, fiber tracking methods were developed. The classical streamline tracking methods cannot handle voxels that contain multiple fiber orientations well. Examples for such multi-orientation voxels are voxels containing fiber crossing, fiber branching or kissing fibers. The directional heterogeneity in such voxels cannot be modeled with the simple second order diffusion tensor used for standard tracking, which can only model one dominating diffusion direction. I have developed a new diffusion simulation based fiber tracking method that uses time-of-arrival maps to reconstruct the tracts. This method is able to resolve a lot of problems of the commonly used streamline method. The reconstruction can also be performed on simulations that use more advanced diffusion models. A first simple example for such an extension of the diffusion simulation is discussed here. To use more advanced diffusion models, such as the higher order tensor hierarchy, in a clinical environment the extensive measurement time required for the data acquisition for this model needs to be reduced. In my evaluations on gradient encoding schemes I investigated the required number of encoding directions and the distribution of the directions for higher order tensor estimations with several quality measures. I was able to show, that 21 directions are the minimal required number of encoding directions for the higher order tensor hierarchy model and that the paired off force minimizing gradient encoding scheme is the best all purpose scheme for diffusion evaluation with second or higher order tensor models