2009 Summer Project Week Quantify VMTK seg error

Quantifying User-Variability Error Using VMTK Level Set Segmentation

Intra-cranial aneurysm rupture is a poorly understood phenomenon. The pathology is not self-presenting, so many patients hemorrhage without diagnosis. When aneurysms are detected serendipitously, clinicians currently make interventional decisions based on the maximum size of the aneurysm. To date, there is strong statistical evidence to suggest that aneurysm size is a poor measure of rupture risk.

Our lab is attempting to assess rupture risk by quantifying the shape of aneurysms, coupling that information with the stress-strain on the vessel walls and the hemodynamics of the flow. One of the critical factors to this research is understanding the variability that user-defined points have in the determination of segmentation of the Circle of Willis and the aneurysm.

Our project proposes to simulate multiple user sessions on a data set, measure the surface variability and determine the 'average' surface for segmentation.