VesSAP

VesSAP segments and quantifies whole mouse brain vasculature from 3D imaging datasets to enable micrometer-scale analysis and anatomical localization.

Key Features:

• Convolutional neural network (CNN): Uses a CNN architecture for segmentation of brain vasculature from volumetric imaging data.
• Transfer learning: Applies transfer learning to improve segmentation accuracy and adaptability across datasets.
• Human-level accuracy: Achieves segmentation performance reported at human-level accuracy for identifying and delineating vascular features.
• Scalability and micrometer-scale processing: Processes whole mouse brains at micrometer resolution to handle large three-dimensional imaging datasets.
• Atlas integration: Registers segmented vascular structures to the Allen Mouse Brain Atlas for anatomical localization and comparison.
• Quantification: Produces unbiased quantitative measures of vascular features across samples and regions.

Scientific Applications:

• Angioarchitecture analysis: Enables comprehensive analysis of cerebral vascular architecture in mouse brains.
• Species- and strain-specific vascularization: Supports comparison of vascular patterns across mouse strains including C57BL/6J, CD1, and BALB/c and revealed secondary intracranial collateral vascularization in CD1 mice.
• Regional vascular comparisons: Facilitates comparisons of vascularization between brain regions, such as demonstrating reduced vascularization in the brainstem relative to the cerebrum.

Methodology:

Applies a deep learning–based segmentation pipeline using a convolutional neural network with transfer learning, followed by registration of segmented vasculature to the Allen Mouse Brain Atlas.