Estimating Perfusion Deficits in Acute Stroke Patients Without Perfusion Imaging

Abstract

Background: Perfusion weighted imaging (PWI) is critical for determining whether stroke patients presenting in an extended time window are candidates for mechanical thrombectomy. However, PWI is not always available. Fluid-attenuated inversion recovery hyperintense vessels (FHVs) are seen in patients with a PWI lesion. We investigated whether a scale measuring the extent FHV could serve as a surrogate for PWI to determine eligibility for thrombectomy.

Methods: The National Institutes of Health (NIH) FHV score was developed to quantify the burden of FHV and applied to magnetic resonance imaging scans of stroke patients with fluid-attenuated inversion recovery and perfusion imaging. The NIH-FHV was combined with the diffusion weighted image volume to estimate the diffusion-perfusion mismatch ratio. Linear regression was used to compare PWI volumes and mismatch ratios with estimates from the NIH-FHV score. Receiver operating characteristic analysis was used to test the ability of the NIH-FHV score to identify a significant mismatch.

Results: There were 101 patients included in the analysis, of whom 78% had a perfusion deficit detected on PWI with a mean lesion volume of 47 (±59) mL. The NIH-FHV score was strongly associated with the PWI lesion volume (P<0.001; R²=0.32; β-coefficient, 0.57). When combined with diffusion weighted image lesion volume, receiver operating characteristic analysis testing the ability to detect a mismatch ratio ≥1.8 using the NIH-FHV score resulted in an area under the curve of 0.94.

Conclusions: The NIH-FHV score provides an estimate of the PWI lesion volume and, when combined with diffusion weighted imaging, may be helpful when trying to determine whether there is a clinically relevant diffusion-perfusion mismatch in situations where perfusion imaging is not available. Further studies are needed to validate this approach.

Keywords: ROC curve; humans; ischemic stroke; linear models; perfusion.

Figure 1:

NIH-FHV score for an MCA perfusion deficit: two slices of the FLAIR sequence of a patient and the corresponding perfusion deficit (dark blue arrows) on PWI are shown. There are not more than two FHV in the frontal region (red arrows) on a single slice thus it contributes 1 point. The insular (yellow) and temporal regions (green arrows) both have greater than two FHV on a single slice and contribute two points each. The parietal region (light blue arrows) contributes one point. The NIH-FHV score for this patient is 6.

Figure 2:

Examples of calculating the NIH-FHV score in the ACA and PCA territories are shown. A NIH-FHV score of 1 (red arrow) is assigned to the ACA PWI deficit (blue arrow). A NIH-FHV score of 2 (red arrows) is assigned to the PCA PWI deficit (two blue arrows).

Figure 3:

Examples from two different patients are show in which FHV are detected on the FLAIR scan (red arrows) in an area of ischemia on DWI despite the lack of a perfusion deficit reaching the 6 second threshold on the TTP map.

Figure 4:

Panel A shows a scatter plot comparing the estimated mismatch (NIH-FHV / DWI volume) with the actual mismatch (PWI volume / DWI volume) for patients with PWI volume greater than 15 mL. Panel B shows the ROC curve for detecting a PWI volume > 15 mL with the NIH-FHV score. Panel C shows the ROC curve for detecting a mismatch ratio > 1.8 using the FHVmr (NIH-FHV score / DWI volume).

Figure 5:

Two slices from both 2-dimentional (left) and 3-dimentional (right) FLAIR sequences acquired during the same scan for a single patient with a perfusion deficit are shown. The 2D FLAIR shows FHV (arrows) due to diminished inflow/outflow of arterial blood moving through the slices being acquired. The 3D FLAIR, for which all slices are acquired simultaneously, does not demonstrate FHV in the same regions (circles).