The Dti Data Reconstruction And Visualization

Diffusion Tensor Imaging (DTI) is a fairly new Magnetic Resonance Imaging technique. It shows microscopic structure of tissue and lesion by analyzing the diffusion of water molecules in them. DTI data reconstruction and visualization can display tract trajectories in tissue, which can be used in non-invasive exploration of the micro structure in living human, diagnosis of neurological diseases and preoperative surgical plan.A visualization tool for processing and visualization DTI data is developed with C++and OpenGL. It implements a variety of visualization technology, for instance, visualization for tensor, anisotropy (FA, VA, MD) maps, primitive representation, multi-planar reconstruction and fiber tracking. Involved algorithms are described in detail, especially fiber tracking algorithms such as FACT, Tensorline and Brute force algorithms. Advantages and disadvantages of different fiber tracking algorithms are summarized. Tensorline and Brute force algorithms are combined to present the tracking results.Furthermore, to achieve real time fiber tracking, we use GPU to speed up fiber tracking by means of CUDA. A new computing model is proposed to make full use of GPU's parrallel ability. Our fiber tracking algorithm's implementation have achieved a 10x performance gain. The proposed model can be easily used to accelerate other fiber tracking algorithms.