Optimization of collateral grading on computer tomography angiography

Abstract

Background

As compared to single-phase CTA (sCTA), multi-phase CTA (mCTA) has been shown to more accurately estimate collateral flow in acute ischemic stroke (AIS). We sought to determine the characterization of poor collaterals across the three different phases of the mCTA. We also attempted to establish the optimal arterio-venous contrast timing parameters on sCTA that would prevent false positive reads of poor collateral status.

Methods

We retrospectively screened consecutive patients admitted for possible thrombectomy from February 2018 to June 2019. Only cases with intracranial internal carotid artery (ICA) or main trunk of the middle cerebral artery (MCA) occlusion and both baseline mCTA and CT Perfusion available were included. Mean Hounsfield units (HU) of torcula and torcula/patent ICA ratio were used for the arterio-venous timing analysis.

Results

Of the 105 patients included, 35 (34%) received IV-tPA treatment and 65 (61.9%) underwent mechanical thrombectomy. A total of 20 patients (19%) had poor collaterals on the third-phase CTA (ground-truth). The first-phase CTA often underestimated collateral score (37/105 [35%], p < 0.01), however there were no significant differences across the second- and third-phases (5/105[5%], p  =  0.06. Venous opacification Youden's J point for identifying suboptimal sCTAs was found to be 207.9HU in the torcula (65% sensitivity,65% specificity) and 66.74% for torcula/patent ICA ratio (51% sensitivity,73% specificity).

Conclusion

A dual-phase CTA is significantly similar to a mCTA assessment of collateral score and may be applied at community-based centers. Absolute or relative thresholds for torcula opacification may be used to identify poor bolus-scan timing thus preventing erroneous assumptions of poor collaterals on sCTA.