Quantitative Analysis and Comparison of Cerebrovasculature in Common Mouse Strains: C57BL/6, CD-1, CBA, and 129/Sv using Imaging, Automatic Segmentation and Labelling of the Cerebral Vessels

The study of cerebral vascular development can broaden our understanding of underlying variations such as pathologies that lead to cerebrovascular disorders. The development of high resolution 3D imaging modalities such as computed tomography angiography, has provided us with a great means of visualizing vasculature anatomy. However, the high complexity and 3D nature of the cerebral vasculature make comparison and analysis difficult, and time-consuming. I present a novel optimization of an imaging method for visualizing the complete cerebrovasculature in mouse models using micro Computed Tomography. I present an algorithm for automated segmentation and recognition of the cerebral vessels of mouse brain. The automatic vessel segmentation method is based on extracting a network of tubular objects from the background using the intensity contrast information of the images. The segmented vasculature is represented as an attributed graph where each vessel segment is an edge in the graph with local attributes such as diameter, length and direction of that vessel, as well as global features such as connectivity of vessel segments. Each vessel is annotated manually with its corresponding anatomical label in ``Brain-view2", an interactive visualization software that I have helped develop, to form a training set. The segmented network of vessels are labelled automatically using Bayesian inference on the training set and the local and global features of the cerebrovasculatures. Currently, no other fully-automated recognition tool for identification of each vessel in the cerebral vascular system is available. I have used this automatic recognition method to study the differences in the cerebrovasculature of four common mouse strains, C57BL/6 (n=10), CD-1 (n=8), CBA (n=9), and 129/Sv (n=8). The known variations in the circle of Willis and the perfusion territories of main cortical arteries were confirmed. I also quantified previously unknown inter-strain variations in the venous circulatory tree and the posterior arterial circulation tree of these four common mouse strains. The automatic recognition of the cerebral vascular system enables us to study the vascular variations among individuals. Also, this tool can be employed to study cerebrovascular development and can contribute to understanding of the vascular adaptation in response to pathology or treatment.Ph.D