Feasibility and diagnostic accuracy of using brain attenuation changes on CT to estimate time of ischemic stroke onset

Grant Mair,Awad Alzahrani,Richard I. Lindley,Peter A. G. Sandercock,Joanna M. Wardlaw

DOI: https://doi.org/10.1007/s00234-020-02591-w

2020-10-30

Neuroradiology

Abstract:Abstract Purpose CT attenuation of ischemic brain reduces with time after stroke onset. We aimed to quantify this relationship and test the feasibility and accuracy of estimating stroke onset time using only CT attenuation of visible ischemic lesions, the CT-Clock Tool . Methods We selected CT scans with ischemic lesions representing a range of stroke-onset-to-scan times (elapsed time) from a well-defined stroke trial. We measured the attenuation of ischemic lesions and contralateral normal brain to derive attenuation ratio. We assigned scans to development (75%) or test (25%) datasets. We plotted the relationship between attenuation ratio and elapsed time in the development dataset and derived a best-fit curve. We calculated estimated time in the test dataset using only the attenuation ratio curve. We compared estimated time to elapsed time and derived absolute error for estimated time. We assessed area under the receiver operating characteristic (AUROC) curve for identifying scans ≤ 4.5 h elapsed time. Results We included 342 scans from 200 patients (41% male, median age 83 years). Elapsed time range: 22 min to 36 days. Estimation errors were least at early elapsed times (r = 0.82, p 3–9, > 9–30 and > 30 h from stroke onset, respectively. AUROC was high at 0.955. Conclusions It is feasible to accurately estimate stroke onset time using simple attenuation measures of ischemic brain. Our method was most accurate 0–9 h from onset and may be useful for treatment eligibility assessment, especially where imaging resources are limited.

radiology, nuclear medicine & medical imaging,clinical neurology,neuroimaging