My PhD research, so far, is primarily focused on Diffusion Tensor Imaging (DTI) and its statistical analysis. To be specific, I have worked on DTI Atlas building and tract based statistical analysis to perform group statistics.
 |
DTI Atlas buildingDTI Atlas building tries to map all the DTI images into a common coordinate framework in order to facilitate statistical analysis of diffusion parameters across the population. The pipeline first estimates the individual tensors and their respective scalar measures using Dtiprocess tools. Next, an unbiased, deformable atlas building procedure is applied by using AtlasWerks to bring the population of datasets into the common coordinate frame. To maintain the relationship between the deformed diffusion images and the respective tensor orientation, we use Riemannian framework to average tensors. For more details about the pipeline, click here |
 |
Tract-based Statistical analysisAfter reconstructing fiber-tracts using streamline integration method, the diffusion measures corresponding to these tracts provide a means to perform population studies in neuroimaging. DTI atlas building produces a global spatial normalization in order to parameterize these diffusion properties across the population. Thus, it provides a clinically interpretable structure to analyze group differences. To carry out these group differences along the tracts, the techniques including Principal Component Analysis (PCA) and nonparametric permutation tests have been used. |
 |
Volumetric Approach to segment WM tract in DTI
A volumetric segmentation framework is used to study the
white matter connectivity, specifically the corticospinal
tract (CST). The methodology is based
on Fletcher et
al and uses Hamilton-Jacobi (H-J) formulation and
a fast iterative method to minimize the cost between two
target regions. The resulting high-connectivity voxels
represent the volumetric pathway of the required tract
between two regions. |
