TY - JOUR
T1 - A benchmarking tool to evaluate computer tomography perfusion infarct core predictions against a DWI standard
AU - Cereda, Carlo W.
AU - Christensen, Soren
AU - Campbell, Bruce C. V.
AU - Mishra, Nishant K.
AU - Mlynash, Michael
AU - Levi, Christopher
AU - Straka, Matus
AU - Wintermark, Max
AU - Bammer, Roland
AU - Albers, Gregory W.
AU - Parsons, Mark W.
AU - Lansberg, Maarten G.
PY - 2016
Y1 - 2016
N2 - Differences in research methodology have hampered the optimization of Computer Tomography Perfusion (CTP) for identification of the ischemic core. We aim to optimize CTP core identification using a novel benchmarking tool. The benchmarking tool consists of an imaging library and a statistical analysis algorithm to evaluate the performance of CTP. The tool was used to optimize and evaluate an in-house developed CTP-software algorithm. Imaging data of 103 acute stroke patients were included in the benchmarking tool. Median time from stroke onset to CT was 185 min (IQR 180-238), and the median time between completion of CT and start of MRI was 36 min (IQR 25-79). Volumetric accuracy of the CTP-ROIs was optimal at an rCBF threshold of <38%; at this threshold, the mean difference was 0.3 ml (SD 19.8 ml), the mean absolute difference was 14.3 (SD 13.7) ml, and CTP was 67% sensitive and 87% specific for identification of DWI positive tissue voxels. The benchmarking tool can play an important role in optimizing CTP software as it provides investigators with a novel method to directly compare the performance of alternative CTP software packages.
AB - Differences in research methodology have hampered the optimization of Computer Tomography Perfusion (CTP) for identification of the ischemic core. We aim to optimize CTP core identification using a novel benchmarking tool. The benchmarking tool consists of an imaging library and a statistical analysis algorithm to evaluate the performance of CTP. The tool was used to optimize and evaluate an in-house developed CTP-software algorithm. Imaging data of 103 acute stroke patients were included in the benchmarking tool. Median time from stroke onset to CT was 185 min (IQR 180-238), and the median time between completion of CT and start of MRI was 36 min (IQR 25-79). Volumetric accuracy of the CTP-ROIs was optimal at an rCBF threshold of <38%; at this threshold, the mean difference was 0.3 ml (SD 19.8 ml), the mean absolute difference was 14.3 (SD 13.7) ml, and CTP was 67% sensitive and 87% specific for identification of DWI positive tissue voxels. The benchmarking tool can play an important role in optimizing CTP software as it provides investigators with a novel method to directly compare the performance of alternative CTP software packages.
UR - https://hdl.handle.net/1959.7/uws:63870
U2 - 10.1177/0271678X15610586
DO - 10.1177/0271678X15610586
M3 - Article
SN - 0271-678X
SN - 1559-7016
VL - 36
SP - 1780
EP - 1789
JO - Journal of Cerebral Blood Flow and Metabolism
JF - Journal of Cerebral Blood Flow and Metabolism
IS - 10
ER -
