PDE denoising of MR diffusion tensor imaging data

In diffusion tensor imaging tracktography, the noise level in diffusion weighted images has direct impact on the accuracy of diffusion tensor orientation which is usually used as local fiber direction. Therefore, removing noise becomes one way to improve the signal to noise ratio without increasing the acquisition time. We demonstrate that vector based diffusion filter with norm constraint has rigid theoretical background and better denoising result on directional unit vector field regulation. A tissue property integrated framework for diffusion filter is also introduced for anatomical information driven filtering. The applications in simulated data and in situ mouse brain data reveal improved brain white matter fiber tracking results.