Regularization of bending and crossing white matter fibers in MRI Q-ball fields

In diffusion magnetic resonance imaging with high-angular-resolution diffusion imaging, a set of techniques has become available that allows better acquisition and representation of multidirectional diffusion profiles, e.g., in voxels with crossing, branching and kissing fibers. The poor spatial resolution and low signal-to-noise ratio of the data, particularly when acquired under clinical conditions, prevent tractography algorithms from reliably reconstructing complex white matter structures. With cone-beam regularization, an intervoxel smoothing approach has been described, which, in this article, is refined and adapted to fibers with subvoxel bending. By introducing the concept of asymmetric orientation distribution functions (aODFs), we are able to sharpen diffusion profiles of bending fibers and estimate subvoxel curvature. We also propose a deterministic fiber-tracking algorithm that exploits the enhanced resolution of aODFs. The approach is evaluated quantitatively and compared with state-of-the-art noise-suppression techniques in a study with a biological diffusion phantom. Moreover, we present results from an in vivo study in which we demonstrate the method's ability to optimize tractography of bending fiber pathways of optic radiation.