Inter- and intraobserver reliability of five MRI sequences in the evaluation of the final volume of cerebral infarct

Abstract

Purpose: To evaluate the reproducibility of fluid attenuated inversion recovery (FLAIR) and four other magnetic resonance imaging (MRI) sequences in the quantitative assessment of final cerebral infarct volume. Materials and Methods: FLAIR, T1-3D, magnetization transfer ratio (MTR)-map, diffusion-weighted trace (DWI)-trace, and apparent diffusion coefficient (ADC)-map, were acquired and measured in 33 patients 30-45 days after onset of a first-ever ischemic stroke. The infarct area was visually detected and manually delineated two times by two readers separately after images and sequences randomization. The reliability was assessed by using an intraclass correlation coefficient (ICC) and its two-sided 95% confidence interval (95% CI). Results: DWI-trace had the best reliability, with an ICC of 0.96 (95% CI = 0.93-0.98). FLAIR had an ICC of 0.86 (95% CI = 0.73-0.93), and a much higher volume. T1-3D, MTR-map and ADC-map had lower reliability or excessive volume values equal to 0 in comparison to DWI-trace. Conclusion: DWI-trace performed within 30th and 45th day following onset of acute ischemic stroke was the most reliable sequence for final infarct volume quantification. This sequence should be added to FLAIR evaluation to strengthen the statistical results of the pharmacological trials and reduce their variability. © 2009 Wiley-Liss, Inc.