Sensitivity of visual and quantitative detection of middle cerebral artery occlusion on non-contrast-enhanced computed tomography

INTRODUCTION: This study aims to investigate diagnostic sensitivity and reliability for the detection of middle cerebral artery occlusion (MCAO) on non-contrast-enhanced computed tomography (NECT) by visual assessment (VA), Hounsfield unit (HU) measurement, calculation of the Hounsfield unit/hematocrit (HU/Hct) ratio, and combination of visual assessment and attenuation measurement (VA + HU).

METHODS: NECT of 18 patients with angiographically proven MCAO and 18 patients without MCAO were reviewed by two blinded observers. Visual assessment of presence or absence of a hyperdense sign was followed by HU measurement of both middle cerebral arteries (MCA). Sensitivity, specificity, positive predictive value, and negative predictive value were calculated for VA, HU measurement, HU/Hct ratio, and VA + HU measurement. Receiver operating characteristic curve analysis (ROC) was performed to determine the optimal cut-off values for MCAO using attenuation measurements or HU/Hct ratio.

RESULTS: Diagnostic sensitivity/specificity was 63%/91% for VA, 56%/88% for attenuation measurement, 68 %/81 % for HU/Hct ratio, and 75%/78% for VA + HU. ROC curve analysis revealed cut-off values of >42.5 HU for attenuation measurements and >1.12 HU/Hct for HU/Hct ratio.

CONCLUSION: Combination of visual assessment with additional attenuation measurement with a cut-off value of 42.5 HU is recommended for most sensitive and reliable detection of MCAO on NECT.