Fiber-tracking through Multiple Sclerosis Lesions Using Probabilistic Tracking

INTRODUCTION Multiple sclerosis (MS) is a condition characterized by progressive demyelination of axons and loss of white matter (WM) integrity in the CNS. Brain injury in MS patients includes both inflammatory and demyelinating changes that range from lesions that do not cause overt loss of function to those that result in complete transection of axons, long-term axonal degeneration, and thus severe functional deficits. Such deficits are pathway-specific insomuch that function in connected brain regions and degree of such connectivity is compromised following disruption of individual WM pathways. It is of interest, therefore, to develop highly sensitive pathway-specific markers of MS disease progression, and use them to evaluate axonal integrity of fibers that propagate through MS lesions. Diffusion tensor imaging (DTI) is widely used for localization of individual WM pathways. Even though DTI methods are sensitive to detection of acute and chronic MS lesions and the lesion burden in normal appearing white matter (NAWM), the quantification of pathway-specific axonal damage depends on the ability of such methods to track the entire length of fibers as they propagate through the lesions. Traditional deterministic methods use the MR signal that encodes the directionality of water diffusion to construct a diffusion tensor, which in turn describes the vector of the preferentially oriented diffusion of water in each location throughout the brain. Once obtained, the direction of such a vector is deterministically integrated along these directions to produce a line. Although this line often mimics the actual directions of axonal fibers, it only maps the direction of maximal diffusion. While this technique is conceptually straightforward and may work well for large fiber bundles in normal brain tissue, it is unstable and often breaks down when it experiences crossing fibers or a region of low anisotropy, leading to inaccurate estimation of tracking, or failure to identify tracks at all. The effect of MS lesions is likely to be similar, since the effect of a lesion is inflammatory in nature increasing isotropy of water diffusion. Deterministic methods are thus poorly suited to track through lesions. In contrast, probabilistic tracking, when combined with non-tensor fiber orientation distribution estimation methods, can avoid such pitfalls. It is hypothesized that methods utilizing probabilistic tracking are capable of propagating tracks through such isotropic anomalies as MS lesions. This study specifically aimed to validate probabilistic tracking as a reliable tool that allows fiber tracing to propagate through MS lesions. Additionally, driven by the overall goal to characterize WM integrity in MS patients, this work has begun applying this method to quantitatively evaluate diffusion parameters of fiber tracks that traverse MS lesions.