Ewoud J. Smit

In acute stroke, CT angiography (CTA) and CT perfusion (CTP) are generally combined for their complementary information. Previously, we have demonstrated that CTA can be derived from CTP data with good image quality. So, we wondered if these derived CTA images could also be used for occlusion detection, and performed a diagnostic performance study.

We found that the derived CTA images provide a diagnostic accuracy similar to that of standard CTA for the detection of artery occlusions in acute stroke. All large vessel occlusions, defined as internal carotid artery, basilar artery, and middle cerebral artery segment 1, were detected on both techniques. Occlusions in smaller arteries were occasionally missed on both techniques, mainly because of observer variation.

While the standard and derived CTA images generally look virtually identical, they may differ in case of contrast delay. Standard CTA is performed at one moment in time, and vessels where contrast arrives later will not be visible. The derived CTA images from CTP data are delay-insensitive because they display maximal contrast enhancement over time (temporal Maximum Intensity Projection, tMIP). Because of this delay insensitivity, we refer to this technique as timing-invariant CTA. Previously, we have demonstrated that the image quality of timing-invariant CTA is drastically improved if the tMIP is preceded by modest temporal filtering, and showed that the image quality is superior to that of standard CTA and 4D-CTA. Other studies have shown that timing-invariant CTA is better than standard CTA in visualizing collateral vessels and the assessment of thrombus length because of the delay-insensitivity.

Our ultimate goal is to develop a single acquisition stroke protocol, where noncontrast CT, timing-invariant CTA, 4D-CTA, and perfusion maps are derived from one dynamic acquisition. In this way, patient safety is enhanced by reducing total radiation dose and contrast material usage. Timing-invariant CTA is currently available in several commercial software packages (with or without temporal filtering). In our clinic, where whole brain coverage is available, timing-invariant CTA is reconstructed for all brain perfusion studies. The images are mainly used for the assessment of collaterals, thrombus length, and pseudo-occlusions due to delay. Timing-invariant CTA has not yet obviated standard CTA in clinical practice, mainly because the cervical arteries are not visualized. To solve this, we are currently investigating a modified brain perfusion protocol with one interleaved acquisition of the neck, for which preliminary results look promising.

Read this article at AJNR.org …