A segmentation-based volumetric approach to localize and quantify cerebral vasospasm based on tomographic imaging data.

Abstract

Quantification of cerebral vasospasm after subarachnoid hemorrhage (SAH) is crucial in animal studies as well as clinical routine. We have developed a method for computer-based volumetric assessment of intracranial blood vessels from cross-sectional imaging data. Here we demonstrate the quantification of vasospasm from micro computed tomography (micro-CT) data in a rodent SAH model and the transferability of the volumetric approach to clinical data.We obtained rodent data by performing an ex vivo micro-CT of murine brains after sham surgery or SAH by endovascular filament perforation on day 3 post hemorrhage. Clinical CT angiography (CTA) was performed for diagnostic reasons unrelated to this study. We digitally reconstructed and segmented intracranial vascular trees, followed by calculating volumes of defined vessel segments by standardized protocols using Amira® software.SAH animals demonstrated significantly smaller vessel diameters compared with sham (MCA: 134.4±26.9μm vs.165.0±18.7μm, p<0.05). We could highlight this difference by analyzing vessel volumes of a defined MCA-ICA segment (SAH: 0.044±0.017μl vs. sham: 0.07±0.006μl, p<0.001). Analysis of clinical CTA data allowed us to detect and volumetrically quantify vasospasm in a series of 5 SAH patients. Vessel diameters from digital reconstructions correlated well with those measured microscopically (rodent data, correlation coefficient 0.8, p<0.001), or angiographically (clinical data, 0.9, p<0.001).Our methodological approach provides accurate anatomical reconstructions of intracranial vessels from cross-sectional imaging data. It allows volumetric assessment of entire vessel segments, hereby highlighting vasospasm-induced changes objectively in a murine SAH model. This method could also be a helpful tool for analysis of clinical CTA.