Three-dimensional Segmentation of Blood Vessels from Intensity In-homogeneous Medical Images

Blood vessel segmentation helps to know the progress of the disease during diagnosis. The presence of intensity in-homogeneity, conglutination tissue and noise in medical images has led to difficulty in extraction of different sizes of blood vessels, difficulty in separating vessels for further analysis (such as quantification of angiogenesis), difficulty in distinguishing vessels from non vessels. Most of the available techniques are 2D-based. Despite the fact that 2D-based segmentation is easy, it doesn't provide full information about anatomic structure. Most of the available 3D vessel segmentation techniques require human intervention and fail to segment different sizes of vessels. In this paper, a 3D hybrid approach for segmentation has been developed, based on white top hat scale space bilateral hessian vessel enhancement filter and hysteresis threshold method combined with multi-threshold Otsu method. The hybrid method can address noise and intensity in-homogeneity problem, as a result, more vessels of different sizes are detected. The method can also incorporate spatial information, abnormalities in the vessels are detected. Vessels of different sizes are separated to ease further analysis. Conglutination tissue (that obstruct segmentation process) is eliminated to ease extraction of different sizes of vessels.