jist:ConnectivityMatrix

From NITRC Wiki

Connectivity Analysis

An example connectivity matrix

Connectivity analysis of the brain produces connectivity matrices. A connectivity matrix is a symmetric matrix describing the connectivity strength between pairs of regions of interest (ROIs). This is computed using three steps.

1. ROI definition (labeling)
2. Path estimation (tractography)
3. Computation of the connectivity measure

Most often, entries in the matrix are the mean of some measure over the path between the ROIs. The path between regions is found using DTI tractography.

Labeling

First the regions of interest (ROIs) that we will estimate the connectivity between must be defined. In our case, we use gyral labels computed on cortical surfaces as our ROIs.

Computing Tracts

Next, the paths through space that connect these regions must be estimated. This is done using DTI data, and fiber tractography.

The Gyral Labeling - Connectivity computation processing pipeline

Connectivity Measures

A number of different measure can be computed with connectivity analysis. These measures fall into two categories:

• Fiber-based measures
  • Properties of the streamlines that define the path of connectivity between ROIs. They include, among

others, mean fiber length and median curvature.

• Volume-based measures
  • Properties of the volume (voxels) that are on the path between ROIs. Specifically, a voxel is part of the path between two regions if it contains a fiber joining the two regions of interest.