Abstract 2251: Signal-To-Noise Ratio and Acquisition Duration Explain Erroneous Mean Transit Times in Acute Stroke Computed Tomography Perfusion Imaging
Introduction: Erroneous, decreased or even negative Mean Transit Time (MTT) estimated by Computed Tomography Perfusion (CTP) imaging in acute stroke hypoperfused areas have been reported. We use CTP simulation data to investigate this issue.
Hypothesis: Two possible hypotheses to explain this phenomenon: voxel-wise CTP Signal-To-Noise Ratios (SNR) are too low and/or concentration-time curves (CTC) are truncated before relaxation to baseline.
Materials: In silico data were prepared to match typical in vivo CTP acute stroke hypoperfused data previously described in the literature. To investigate the impact of truncation, acquisition duration was set to D=40, 50, 60, 70, 80 and 160s. To investigate the impact of SNR, SNR was set to 10, 20, 40, 80 and 160. Data were processed by singular values decomposition with oscillation index (oSVD) (POI=0.095) and a Bayesian reference method (BP).
Results: For low Cerebral Blood Volumes CBV<=3%, simulations with in vivo parameters reproduce Cerebral Blood Flow (CBF) and MTT empirical distributions well: oSVD returns negative MTT while BP returns increased MTT most of the time, oSVD MTT are negatively correlated while BP MTT are positively correlated with true values (see figure). Increasing D degrades the estimates even more. Increasing SNR improves low MTT<=15s estimates if SNR>=40 for BP and SNR>=80 for oSVD. At D=50s, high MTT>=15s are always underestimated due to CTC truncation but positively correlated with BP and negatively correlated with oSVD. At high SNR, low CBF estimates are excellent with BP but have poor linearity with oSVD.
Conclusion: CTP SNR are too low to reliably estimate CBF and MTT in acute stroke hypoperfused areas. Spatial downsampling, averaging or regularization is required to reach higher SNR. The Bayesian method is more robust against noise than oSVD. Short acquisition duration prevents reliable estimation of high MTT, which can potentially impact on stroke assessment in clinical practice.
- © 2012 by American Heart Association, Inc.