Purpose Patient selection for endovascular revascularization based on penumbral imaging may help to improve outcomes for victims of acute ischemic stroke. Although MR perfusion–diffusion mismatch is well validated, its clinical application has been hampered by lack of availability of MR imaging in the ER setting. Increasingly, CT based perfusion–cerebral blood volume mismatch is employed to triage patients. The goal of this work is to compare CT cerebral blood volume (CBV), MR diffusion weighted imaging (DWI) and histopathology in an experimental model of acute ischemic stroke.

Materials and methods Seven purpose bred adult beagle dogs (mean weight 10 kg) were anesthetized as per the procedures approved by our IACUC. Autologous blood clot was prepared by mixing whole blood with thrombin and barium, and allowed to age for 24 h. Following baseline MR imaging, a 5 F catheter was placed into the selected internal carotid artery and the clot fragment (mean length 1 cm) was delivered into the middle cerebral artery. MRI was performed on a 3.0 T system using an eight channel receive only SENSE knee coil and serial diffusion (prestroke and 0.5, 1, 2.5 and 4 h post stroke) and perfusion (1.5 and 4 h post stroke) sequences were acquired. Immediately following the last diffusion scan, the animals were transferred to the adjacent angiographic suite where non-contrast and contrast enhanced flat panel cone beam CT sequences were acquired. Animals were then killed and the coronal brain sections were stained with 2% 2,3,5-triphenyltetrazolium chloride (TTC).

Apparent diffusion coefficient (ADC) maps were generated and imported into Matlab for image processing. ADC values below 0.53×10⁻³ mm²/s were identified and segmented. Using a calibration curve, the CBCT data were converted into Hounsfield units (HU). CBV was determined by ΔHU_Brain/ΔHU_Blood×V_Voxel×N, where the change was between the contrast and non-contrast studies, V_Voxel was the volume of the voxel and N was the number of voxels in 100 g of brain. At this time, CBV maps were generated and the threshold for abnormal CBV was set at 1.0 ml/100 g of brain (based on the canine data). Volumes were derived by taking the product of the segmented area and the slice thickness. Time to peak perfusion images were analyzed using commercial software.

Results TIMI grade 0 flow was successfully induced with an embolus lodged in the middle cerebral artery in all animals. The perfusion imaging revealed hypoperfused brain with heterogeneous severity due to the variability of collateral circulation. The final infarct size, as determined by the ADC decrease, was 4385±1930 mm³ (mean±SE), approached the mean MR perfusion lesion (6483±1515 mm³) and was found comparable with lesion volume measured histologically using TTC (4408±1805 mm³). The mean CBV lesion from CT slightly overestimated (5199±741 mm³) the infarct size. Compared with the very close agreement between DWI and TTC (R²=0.92), the correlation between the CBV lesion volume and TTC was not as strong (R²=0.78).

Conclusion Our preliminary results indicate that while there is a close agreement between infarct lesion volumes measured on MR DWI and histopathology, estimates of lesion volume from CT CBV requires further study.