Automatic Segmentation and Quantification of TB Scale Volumetric Murine Brain Vasculature Data

Emerging serial section light microscopy platforms, like Knife-Edge Scanning Microscopy (KESM), generate high-resolution data sets at a rate exceeding 1 TB /cm3 of tissue, and can generate 3D voxel data of an entire mouse organ like a brain or kidney. Not only is this technique much faster than imaging slides manually on a traditional microtome it generates much larger and better statistically sampled data sets. These large datasets require new and innovative infrastructure to support the development and deployment of automated segmentation algorithms. In this paper we briefly describe the KESM, our analysis infrastructure, and our validation procedures for automated tissue segmentation routines. We demonstrate multiparametric quantification of vasculature across large sample volumes and the validation of segmentation of these volumes using both trained pathologists and un-trained workers. This type of validated vascular analysis is useful for understanding tumor angiogenesis, arteriosclerosis, vasculopathies, and neurodegenerative diseases. To that end, recent studies have shown an increase in blood vessel density & reduction in blood vessel diameter in the striatum of mice with Huntington's Disease [1,2]. We show that our validated mouse model recapitulate and expand on these findings across larger volumes.