Objectives: We compared a multi-echo gradient-echo magnetic resonance sequence (susceptibility-weighted angiography [SWAN]) with the T2* sequence for the detection of an arterial thrombus in acute ischaemic stroke.
Methods: Seventy-four consecutive patients with acute ischaemic stroke were included. Proximal arterial occlusions were diagnosed using time-of-flight (TOF) magnetic resonance angiography (MRA). Two-dimensional (2D) axial reformats from 3D SWAN were generated to match with 2D T2* images. For arterial thrombus detection, each set of MR images (T2*, 2D SWAN reformats and 3D multiplanar SWAN images) was examined independently and separately by three observers who assigned the images to one of three categories: (0) absence of thrombus, (1) uncertain thrombus, (2) certain thrombus. Agreement and diagnostic accuracy were calculated.
Results: Twenty-four proximal arterial occlusions involving the anterior (n = 20) or posterior (n = 4) circulation were found. Inter-observer agreement was moderate using T2* images (κ = 0.58), good using 2D SWAN reformats (κ = 0.83) and excellent using multiplanar SWAN images (κ = 0.90). For the diagnosis of thrombus, T2* images were 54% sensitive and 86% specific, 2D SWAN reformats were 83% sensitive and 94% specific and SWAN multiplanar analysis was 96% sensitive and 100% specific.
Conclusions: Three-dimensional SWAN sequence improves the detection of arterial thrombus in patients with acute ischaemic stroke in comparison with the 2D T2* sequence.
Key points: • Multi-echo gradient-echo MR (e.g. susceptibility-weighted angiograph, [SWAN]) is increasingly used in neuroradiology. • Compared with conventional T2* sequences, SWAN improves detection of arterial thrombus. • Multiplanar SWAN analysis had the best diagnostic performance for arterial thrombus detection. • Sensitivity was 96% and specificity 100%. • Findings support combination of time-of-flight and susceptibility effects in suspected acute stroke.