Visual rating of T2′-blood-oxygen-level-dependent magnetic resonance imaging in acute stroke patients -: A pilot study

Purpose: Delineation of brain tissue that is at risk but not yet infarcted (penumbra) continues to be a major challenge for stroke imaging. Metabolic characterization of the penumbra might be able to be achieved using blood-oxygen-level-dependent (BOLD) imaging. Materials and Methods: We analyzed MRI data from 20 patients within the first 6 hours after stroke onset and after 5 - 8 days. Among other sequences, the MRI protocol consisted of diffusion-weighted (DWI/ADC = apparent diffusion coefficient) and quantitative T2 and T2* imaging (qT2, qT2*). BOLD images (T2') were calculated using 1/T2' = 1/qT2* - 1/qT2. BOLD lesions were rated by two blinded observers. Results: Based on the primary blinded reading of the BOLD images, the lesion side was rated correctly by observers 1 and 2 in 80/50% of the cases, incorrectly in 5/40% of the cases, and rated as not visible in 15/10% of the cases. After unblinding the observers, the visibility was rated in 45/45% of the cases as good, in 35/40% of the cases as reasonable, and in 20/15% of the cases as insufficient for diagnostic purposes. The sensitivity for subsequent infarct growth was 0.88 (95% confidence interval, Cl 0.47 to 0.99), the specificity was 0.33 (95% Cl 0.07 to 0.70), the positive predictive value (PPV) was 0.54 (95% Cl 0.25 to 0.81), and the negative predictive value (NPV) was 0.75 (95% Cl 0.19 to 0.99). The odds ratio for subsequent infarct growth was 3.5 (95% Cl 0.20 to 115.53). Conclusion: Hypointense lesions in BOLD imaging were visible and exceeded the lesion in diffusion-weighted imaging in most of the stroke patients. The encouraging results justify further testing of the hypothesis that BOLD lesions, when larger than DWI lesions, are associated with infarct growth from initial DWI to final infarct.