Investigation of Bi-Gaussian kernel for vessel detection in level-set based segmentation framework

Accuracy in segmenting of brain vessels from medical angiographic data is crucial for further modelling and assessment of the human vasculature. It was demonstrated that the level set (LS) approach enhanced by the implementation of the vesselness function (VF) provides a robust segmentation framework enabling the high-quality vessel network extracted from CT and MR images. This work investigates the application of bi-Gaussian kernel for evaluation of VF to further improve segmentation results. Modified VF together with LS was applied for both artificial and real CT data. Obtained sample segmentation results were presented along with discussion and conclusion.