Performance of different MRI based parameter maps to assess penumbral flow in stroke: a 15O-water PET study

Background: The accuracy of perfusion weighted magnet resonance imaging (PW-MRI) based quantitative maps of cerebral blood flow (CBF-MRI), cerebral blood volume (CBV-MRI), mean transit time (MTT-MRI) and relative maps of time to peak (rTTP-MRI) remains a matter of debate. We validated CBF-, CBV-, MTT- and rTTP-PWI on quantitative CBF measurement by 15O-water positron emission tomography (CBF-PET) with respect to penumbral flow (<20ml/100g/min).

Methods: PW-MRI was performed on a 1.5 Tesla scanner (Philips Intera Master) using a 3D EPI-PRESTO sequence; CBF-PET was performed on an ECAT EXACT HR scanner (Siemens/CTI). In acute and subacute stroke patients, the CBF-, CBV-, MTT- and rTTP-PWI and CBF-PET maps were compared on a region of interest based approach. In a receiver operating characteristic (ROC) analysis, the performance of PW derived maps was assessed using quantitative CBF-PET maps with respect to penumbral flow. The best PW threshold to detect penumbral flow as well as its sensitivity and specificity were calculated for each imaging modality.

Results: On visual inspection, an excellent qualitative congruence was found for PWI derived maps. In a pooled analysis of 12 stroke patients (median time MRI to PET: 62 minutes; patients imaged within 20 hours after stroke) the best cut-off value to identify penumbral flow was 20.3ml/100g/min on CBF-MRI, 3.4ml/100g on CBV-MRI, 5.2 seconds on MTT-MRI and 4.3 seconds on rTTP-MRI maps. CBF-MRI and rTTP-MRI showed the highest sensitivity and specificity (CBF: 91% and 90%; TTP: 92% and 89%) for the definition of penumbral flow.

Discussion: Among the commonly used PW parameter maps, quantitative CBF and rTTP maps showed the best estimate of penumbral flow with a sensitivity above 90% and a specificity above 80%. Our results support the MRI derived perfusion measurement in clinical stroke imaging.