Fluid-attenuated inversion recovery (FLAIR) sequences for the assessment of acute stroke : Inter observer and inter technique reproducibility

Diffusion-weighted magnetic resonance (MR) imaging (DWI), and three-dimensional (3D) time-of-flight (TOF) MR angiography (MRA), are highly sensitive for the early detection of stroke and arterial occlusion. However, only a few studies have evaluated the sensitivity of conventional MR sequences that are usually included in the imaging protocol. The aim of this study was to evaluate interobserver and intertechnique reproducibility of Fluid-Attenuated Inversion Recovery (FLAIR) sequences for the diagnosis of early brain ischemia and arterial occlusion. Over a 30-month period, brain MR examinations were performed in 34 patients within 12 hours after stroke onset. Imaging protocol included FLAIR sequences, DWI and 3D TOF MRA. Ten observers including radiologists and neurologists, performed separately a visual interpretation of FLAIR images for the detection of brain ischemia and arterial occlusion seen as an arterial high signal. DWI and 3D TOF MRA were used as reference and interpreted independently by two senior radiologists. Interobserver agreement was assessed for image quality, detectability and conspicuity of lesions whereas intertechnique agreement was only judged for lesion detectability. On FLAIR sequences, interobserver agreement for the detection of brain ischemia and arterial occlusion was excellent (kappa = 0.81 and 0.87 respectively). The concordance between FLAIR and DWI sequences for the detection of brain ischemia and between FLAIR and 3D TOF MRA for the detection of arterial occlusion were judged as excellent for all observers (kappa = 0.91 and 0.89 respectively). Although DWI is the most sensitive technique with which to detect acute stroke, FLAIR imaging may also be useful to demonstrate both acute ischemia and arterial occlusion with an excellent interobserver reproducibility.