(Wednesday, 7/15/2020) 1:00 PM - 2:00 PM [Eastern Time (GMT-4)]
Room: Track 1
Purpose: A cerebral CT perfusion (CTP) scan is almost always performed together with a non-contrast head CT (NCCT) scan during a clinical CT exam for evaluating acute strokes. The purpose of this work is to present a method that utilizes the high quality NCCT images to reduce the noise of CTP baseline images to improve the detectability of ischemic core in parametric perfusion maps.
Methods: Through a cascaded systems analysis of the CTP imaging chain, we theoretically demonstrated that for both deconvolution and nondeconvolution-based CTP systems, the noise of perfusion maps such as cerebral blood volume (CBV) is profoundly influenced by the baseline image noise. Consequently, baseline noise reduction is extremely effective in improving the CNR and detectability of ischemic cores in CBV maps. An optimal weighting scheme was derived for fusing the baseline image frames of the CTP scan with NCCT images acquired with a much higher exposure per rotation. The new composite baseline image was used for the calculation of CBV. This method was evaluated using five in vivo canine subjects with acute ischemic strokes created using an endovascular approach. NCCT and CTP scans were performed using a 64-slice MDCT. Each subject also received a diffusion-weighted imaging (DWI) scan using a state-of-the-art 3T-MRI to generate the gold standard for the location and volume of the infarct core.
Results: The detectability index of the infarct core was improved by 70±18% and the average percent error of ischemic core volume quantification was reduced from 60±2% to 8±6% by using the proposed method.
Conclusion: By leveraging the high quality and freely available NCCT images to reduce CTP baseline image noise, the detectability and quantification accuracy of ischemic infarction core in CBV maps can be effectively improved. The proposed method can be readily integrated into the existing clinical workflow with minimal interruption.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
