Ultra High-Resolution In vivo Computed Tomography Imaging of Mouse Cerebrovasculature Using a Long Circulating Blood Pool Contrast Agent

Abnormalities in the cerebrovascular system play a central role in many neurologic diseases. The on-going expansion of rodent models of human cerebrovascular diseases and the need to use these models to understand disease progression and treatment has amplified the need for reproducible non-invasive imaging methods for high-resolution visualization of the complete cerebral vasculature. In this study, we present methods for in vivo high-resolution (19 μm isotropic) computed tomography imaging of complete mouse brain vasculature. This technique enabled 3D visualization of large cerebrovascular networks, including the Circle of Willis. Blood vessels as small as 40 μm were clearly delineated. ACTA2 mutations in humans cause cerebrovascular defects, including abnormally straightened arteries and a moyamoya-like arteriopathy characterized by bilateral narrowing of the internal carotid artery and stenosis of many large arteries. In vivo imaging studies performed in a mouse model of Acta2 mutations demonstrated the utility of this method for studying vascular morphometric changes that are practically impossible to identify using current histological methods. Specifically, the technique demonstrated changes in the width of the Circle of Willis, straightening of cerebral arteries and arterial stenoses. We believe the use of imaging methods described here will contribute substantially to the study of rodent cerebrovasculature.

: (A) 3D drawings of the bite bar and mouse head holder used for positioning of the animal in the micro--CT scanner. Both of the objects were 3D printed as per the details provided in the materials and method section. (B) Sagittal view demonstrating the supine positioning of the animal during CT image acquisition, insuring airway patency. The circular objects at the bottom are cylindrical iodine--containing phantoms placed in the field of view for image quality control.

Figure 2.
Effect of head holder on animal motion. The animal displays motion (yellow arrows) in the absence of head holder. Based on analysis of skull shift, the motion was determined to be 380 μm. The use of head holder helps minimize such motion. The airways are seen open when the animal is positioned in the head holder. Figure 3: Placement of regions of interest (ROI) for determination of CT attenuation in the (A) internal carotid artery (ICA), and (B) great cerebral vein of Galen (GVG). Background ROI was drawn for determination of noise level, defined as standard deviation in the brain parenchyma, for use in the estimation of vessel SNR.

Statistical Analysis
Analysis of statistical significance was done by first testing for normal distribution of the data. The Kolmogorov-Smirnov non--parametric statistic was used to evaluate normality hypothesis of data distribution. P--values were calculated with a two--sided t--test if the data were normally distributed. Otherwise, statistical analysis was conducted using the two--sided Wilcoxon Rank--Sum test. The effect of protocol parameters i.e., peak voltage (kVp), voxel size (spatial resolution) and contrast agent dose on calculated values of mean CT signal with region of interests, vessel diameters and visibility score were examined. The following hypotheses were tested: a) Effect of spatial resolution (Res): high--resolution 19 µm (HighRes) and low-resolution 35 µm (Low--Res) on vessel diameter assessment (d).
H0:    Figure 7: Effect of spatial resolution (voxel size) and projection number on image profiles of transverse hippocampal arteries. Axial images are presented as 1 mm MIP images acquired at 50 kVp with a contrast agent dose of 2.2 gm I/kg.
Additional results from statistical analysis of data presented in figures in the main article Table 3. Statistical analysis of CT attenuation data presented in Figure 1A of main article. Entries in bold text represent statistical significance (p--value < 0.05). Table 4. Statistical analysis of data on effect of projection number on vessel visibility presented in Figure 4D

Vascular morphometric analysis of ACTA2 knock--out mouse models
The following vascular morphometric parameters were calculated from the cerebrovascular CT images acquired in mouse models of moyamoya disease: • Radius (mm): Median value calculated from a set of minimal radius of segments, (all segments combined together represent the vessel under examination). • Length (mm): Length of centerline of a blood vessel.
• Middle curvature (mm --1 ): For a centerline C and a middle point P on it, there is a unique circle or line that most closely approximates the curve near P. Curvature is expressed as the inverse of the circle radius. • Arc:length Ratio: Also referred to as tortuosity, is calculated as the length along centerline divided by Euclidean distance between beginning and end of centerline. • Narrowing (%): Number of segments that are smaller than 85% of the median radius value. • Mean: Mean aspect ratio is defined as a ratio between first and second eigenvalues calculated for contour for one point along centerline. Mean is calculated for a set of separated aspect ratios. Table 5. Statistical analysis on vascular morphometric parameters of cerebrovascular arteries determined by CT imaging. In all cases, null hypotheses H0: WT = Acta2 +/--, H0: WT = Acta2 --/--, H0: Acta2 +/--= Acta2 --/-were tested. The two--sided Wilcoxon Rank--Sum test was used to assess statistical significance. Bolded entries indicate measures where the null hypothesis was disproven at the 95% confidence level.   Table 6. Statistical analysis of the dimensions in the Circle of Willis determined by CT imaging. For the total area, upper half area and width, the null hypothesis H0: WT = Acta2 +/--, H0: WT = Acta2 --/--, H0: Acta2 +/--= Acta2 --/-were tested. In all cases, the two-sided Wilcoxon Rank--Sum test was used to assess statistical significance. Bolded entries indicate measures where the null hypothesis was disproven at the 95% confidence level.