Thin section magnetic resonance diffusion-weighted imaging in the detection of acute infratentorial stroke.

Abstract

Magnetic resonance diffusion-weighted imaging (DWI) is the most accurate technique available for demonstrating acute infarction; however, false-negative DWI is higher in the infratentorium due to the limited spatial resolution with conventional 5mm DWI. The aim of this study was to compare 5mm DWI with 3mm DWI in the detection of acute infratentorial infarction. A 3mm DWI sequence of the infratentorium was incorporated into the conventional MRI stroke protocol for the evaluation of patients with vertebrobasilar stroke-like deficits. The 5mm and 3mm DWI sequences were assessed by two neuroradiologists who were blinded to the clinical findings. Sensitivity and specificity analysis was then performed against the final clinical diagnosis. The sensitivity for detection of infratentorial infarction was 81.1% for 5mm DWI and 94.6% for 3mm DWI and the specificity was 100% for 5mm DWI and 97.7% for 3-mm DWI. The false-negative rate in detection of infratentorial infarcts was 5.6% for the 5-mm sequence and 1.6% for the 3-mm sequence. The six 5-mm DWI false-negative cases (4.8%) were less than 9mm in diameter (3-8mm, average 4.67mm) and located in the brainstem. This supports the hypothesis that small lesions may not be detected on 5mm DWI due to partial volume averaging. Where there is clinical suspicion of infratentorial infarction, 3mm DWI of the infratentorium adds sensitivity compared to 5mm DWI with only a small reduction in specificity.