Evaluation of the robustness of BrainVISA software on brain sulcus extraction

It is very valuable for early disease diagnosis to analyze cortical surface and sulcus morphology using neuroimaging techniques. There are some platforms that provide automatic segmentation and reconstruction of cortical surface based on MRI data. BrainVISA enables the shape analysis of the sulcal pattern as well as morphological measure of the cortex and has been widely used in many studies. However, there has been no special evaluation to validate this software on populations with different ages. A robust evaluation on automatic sulcus extracting was performed for BrainVISA in this paper. MR Images of 100 healthy subjects were stratified and randomly sampled by age and a sulcus extraction pipeline was defined to operate on these images. BrainVISA showed a quite robust performance with a successful rate of 88%, and there was no significant bias on age or gender (p<;;0.01). The results indicated that BrainVISA could be used reliably on a large extension. Additional experiments showed that BrainVISA would perform better on skull stripped images.