Combined Evaluation of Noncontrast CT ASPECTS and CT Angiography Collaterals Improves Detection of Large Infarcts in Proximal Artery Occlusive Stroke

Farhad Mehrkhani, Olvert A. Berkhemer, Charles B. L. M. Majoie, Mohammad Mansouri, Zahra Karimi, Michael H. Lev, Albert J. Yoo

Research output: Contribution to journalArticleAcademicpeer-review

4 Citations (Scopus)

Abstract

BACKGROUND AND PURPOSE: Imaging may identify patients with very large infarcts who are unlikely to benefit from intra-arterial therapy. Although computed tomography (CT) is widely used, it suffers from poor sensitivity. We sought to evaluate whether combined evaluation of noncontrast CT (NCCT) and CT angiography (CTA) collaterals would improve the detection of large infarcts. METHODS: All patients with anterior circulation proximal artery occlusion and baseline CT, CTA, and magnetic resonance imaging (MRI) performed were identified. NCCT ASPECTS, CTA collateral score (CS), and diffusion-weighted imaging (DWI) lesion volume were determined. Receiver-operating characteristic analyses were performed to test the discrimination of NCCT ASPECTS 0-4, CTA malignant collaterals (CS = 0: absent collaterals in >50% of M2 territory), and the combination for DWI volume > 100 mL. RESULTS: Among 54 patients, mean age was 67 years; median NIHSS was 14. Occlusion locations were ICA terminus (18 [33%]), MCA M1 (20 [37%]), and M2 (16 [30%]). Median NCCT ASPECTS was 8; 8 (15%) had ASPECTS 0-4. Median CTA CS was 2; 9 (17%) were categorized as malignant. Median DWI lesion volume was 25 mL; 12 (22%) had lesions >100 mL. Individually, the CTA malignant collateral profile (98%) and NCCT ASPECTS 0-4 (100%) demonstrated high specificity for DWI lesion volume >100 mL, but had suboptimal sensitivity (both 67%). In the combined approach (CTA CS = 0 and/or NCCT ASPECTS ≤4), the sensitivity improved significantly to 92%, while maintaining high specificity (98%). CONCLUSIONS: Combined evaluation of NCCT ASPECTS and CTA collaterals identifies patients with infarcts >100 mL with high accuracy, and can improve patient selection using current CT techniques.
Original languageEnglish
Pages (from-to)524-529
JournalJournal of neuroimaging
Volume28
Issue number5
DOIs
Publication statusPublished - 2018

Cite this