Research

Characterizing Vascular Connectivity from MicroCT Images

Vascular measurement from noninvasive imaging is important for the study and quantification of vessel disease and can aid in diagnosis, as well as measure disease progression and response to therapy. The analysis of tracked vessel trajectories enables the derivation of vessel connectivity information, length between vessel junction as well as level of ramification, contributing to a quantitative analysis of vessel architecture.

Estimation of Anatomical Connectivity in the White Matter

Diffusion Magnetic Resonance Imaging allows one to capture the restricted diffusion of water molecules in fibrous tissues which can be used to infer their structural organization. In particular, we study level set methods for estimating fiber pathways and determine connectivity likelihood between points in the white matter. We investigate solutions using both Diffusion Tensor Imaging (DTI) and High Angular Resolution Diffusion Imaging (HARDI).

White Matter Parcellation

Once fiber pathways have been extracted, it is desirable to classify them into distinct anatomical structures. This is often referred as white matter parcellation or simply "fiber bundling". The parcellation into distinct tracts enables the quantification of physical as well as geometric properties of fiber bundles. These properties help characterize the white matter and may provide diagnostic information for brain disease and aid in the study of white matter development.

While DTI has shown success in depicting the underlying tissue structure by using a second-order tensor, it cannot represent regions containing more than one fiber population, which is the case in fiber crossings and branches. High Angular Resolution Diffusion images, in which gradients are applied in a large number of directions, attempt to correctly characterize such regions. We exploit different methods to model and better understand the diffusion complexity in areas of fiber variation.

DW-MR image acquisitions are inherently noisy. Classical Gaussian kernels smear out the orientation information embedded in such fields, a fundamental feature used in determining connectivity. We investigate smoothing methods that account for the diffusion profile anisotropy and preserve this directionality information.