Shape Analysis Research Theme

The shape analysis research theme encompasses projects that use geometrical features to increase the meaningfullness and statistical power of morphological and functional imaging analyses. In human brain mapping research, hypothesis testing typically involves obtaining anatomical, functional and diffusion imaging data from multiple subjects, and the analysis must combine data from different subjects in a meaningful way. Shape, considered globally at the level of the entire brain, or locally at the level of a specific structure of interest, can offer meaningful features for statistical hypothesis testing and classification, and shape can also offer a guideline by which to combine multimodal information from multiple subjects.

Shape Analysis using Medial Features

One of the approaches adopted by our group is to fit shape-based coordinate systems to the anatomical structures of interest and to use these coordinate systems to combine information across different subjects. In our work, coordinate systems are fitted to structures by deformable model that has a unique property: the geometric relationship between the boundary and the skeleton (medial axis) of the model is explicitly defined and preserved during deformation. We have applied this shape-based coordinate parameterization to guide group analysis of fMRI and DTI data, finding that it can lead to increased sensitivity and specificity.