Visualization of Neural DTI Vector Fields Using Line Integral Convolution

Diffusion Tensor Imaging (DTI) provides voxel-wise information related to the local diffusion anisotropy. Recent research efforts have centered around the use of this information to infer the direction of local fiber bundles. Calculation of the diffusion tensor and corresponding principle diffusion direction voxel-wise throughout the imaged volume permits the use of tracking algorithms to reconstruct the fiber-bundle pathways. These algorithms are typically based on line propagation and provide results that visually resemble anatomical fiber dissections. Despite the success of these methods, they suffer several limitations, particularly within regions of decreased anisotropy, such as areas of grey matter and fiber crossing or branching. In this paper we present an alternative method of DTI data visualization, line integral convolution (LIC), which has several advantages over existing techniques, particularly, the ability to deal with noise and singularities within the vector field and areas of low anisotropy.