Spectral CTA-based quantitative differentiation of infarcted and non-infarcted ASPECTS regions in acute stroke

Spectral computed tomography (CT) enables material decomposition and monoenergetic imaging, improving infarct-estimation on non-contrast cerebral CT (NCCT). However, beyond NCCT, infarct estimation is also accessible from CT angiography (CTA). We evaluated the potential of spectral CTA to assess quantitative differences of infarcted parenchyma in patients with stroke.

In 14 patients with acute middle cerebral artery stroke, the ten Alberta stroke program early computed tomography score (ASPECTS) regions of the affected and non-affected sides were segmented on NCCT, CTA, CT perfusion-derived cerebral blood flow (CBF) map, and CTA-based spectral maps: virtual-non-contrast, virtual-monoenergetic 40-keV and 100-keV, iodine-density and electron-density. For each region, a ratio to the corresponding contralateral region was calculated. Classification into infarcted and non-infarcted was based on post-thrombectomy follow-up NCCT. On patient-level the groups were summarised with mean ratio-values for infarcted and non-infarcted parenchyma per patient. Conventional images and spectral maps were compared to differentiate the groups.

Except for NCCT, electron-density and virtual-non-contrast maps, there was a significant difference between the ratio-values of infarcted and non-infarcted regions in each spectral map, as well as in CTA and CBF (p < 0.001). Best differentiation was possible with iodine-density (area under the receiving operating characteristic curve [AUROC] = 1.000) and virtual-monoenergetic 40-keV (AUROC = 0.985) maps compared to NCCT, CTA, CBF, electron-density, and VNC maps (AUROC = 0.765‒0.938).

Spectral CTA maps, like virtual-monoenergetic 40-keV and iodine-density, enabled valid quantitative ASPECTS-based infarct-estimation in stroke compared to conventional NCCT, CTA, and CBF. They harbour potential for improving stroke imaging diagnostics.

Quantitative values derived from ASPECTS regions in spectral CTA maps improve differentiation of infarcted and non-infarcted regions compared to conventional images. This technique, therefore, represents an interesting approach to improve diagnostic accuracy in stroke and could potentially assist in making treatment decisions for patients with acute stroke.

In patients with acute stroke, spectral CT improves infarct estimation by NCCT, while its benefit with CTA remains unclear. Quantitative parameters derived from ASPECTS-labelled region of interest were compared between spectral CTA maps and conventional NCCT, CTA, and CBF. Quantitative parameters from virtual-monoenergetic 40-keV and iodine-density maps reliably differentiate infarcted versus non-infarcted ASPECTS regions compared to conventional images.