Introduction: CT Perfusion (CTP) derived ischemia-time dependent thresholds may help with treatment decision making by allowing infarct delineation at admission and prediction of infarct growth over time. The accepted 3h ischemia-time cerebral blood flow (CBF) infarction threshold had been determined by Jones et al in primates using hydrogen clearance. The same threshold from CTP has not been determined. We used a porcine acute stroke model to find this threshold with contemporaneous CTP and PET imaging.
Hypothesis: CTP derived 3h ischemia-time infarction CBF threshold will be near 12 mL100g-1min-1 as determined in primates.
Methods: Cerebral ischemia was induced in the left hemisphere of 11 pigs by injecting endothelin-1 (ET-1) into the cortex. CTP scans were acquired subsequently at 10 min and then 30 min intervals throughout the experiment. At 2.5h, 18F-FFMZ was injected, PET acquisition started at 3h. A median CBF map was produced after co-registering maps from each CTP study. Median CBF maps, PET images, average images from the baseline CTP study, and blood volume (BV) maps from the 10min post ET-1 CTP study were co-registered. ROIs were drawn over the cortex on the affected and contralateral side and superimposed onto all images. Infarct was identified on PET images with a threshold derived from the contralateral ROI. Average images were used to remove white matter from the affected side ROI. Blood vessels were excluded on BV maps with a threshold derived from the affected side ROI. Remaining infarct and non-infarct ROIs were superimposed onto the median CBF map and pixel values were imported into Matlab for ROC analysis.
Results: Of 11 animals, 6 developed infarction. The optimal CBF infarction thresholds determined from the ROC curves were 10.1, 8.9, 15.0, 13.2, 12.1, and 16.2mL100g-1min-1. The average threshold was 12.6 ± 2.8 mL100g-1min-1.
Conclusions: CTP derived CBF threshold for infarction at 3-h ischemia time agrees well with the accepted literature value. Infarction thresholds at other ischemia times can be found using this model by varying the time between ET-1 injection and PET imaging.