Figure 1.

An example of a completed Pipeline workflow (Local Shape Analysis) representing an end-to-end computational solution to a specific brain mapping problem. This pipeline protocol starts with the raw magnetic resonance imaging data for 2 cohorts (11 Alzheimer's disease patients and 10 age-matched normal controls). For each subject, the workflow automatically extracts a region of interest (left superior frontal gyrus, LSFG. using BrainParser [1]) and generates a 2D shape manifold model of the regional boundary [2,3]. Then the pipeline computes a mean LSFG shape using the normal subjects LSFG shapes, coregisters the LSFG shapes of all subjects to the mean (atlas) LSFG shape, and maps the locations of the statistically significant differences of the 3D displacement vector fields between the 2 cohorts. The insert images illustrate the mean LSFG shape (top-right), the LSFG for one subject (bottom-left), and the between-group statistical mapping results overlaid on the mean LSFG shape (bottom-right), red color indicates p-value < 0.01.

Dinov et al. BMC Bioinformatics 2011 12:304   doi:10.1186/1471-2105-12-304
Download authors' original image