Despite this disadvantage, ASL estimates of CBF appeared to be of sufficient quality to localize regions of highest vascularity when compared with DSC. A statistically significant, positive linear correlation was observed between ASL and DSC estimates of mean normalized CBF within both FLAIR hyperintense (Pearson’s correlation coefficient, R2 = .706, P < .0001) and contrast-enhancing regions (Pearson's correlation coefficient, R2 = .809, P < .0001) on a per patient basis (Figs 2A, B). The selleck linear slope that best explained the correlation between normalized

ASL and DSC estimates of CBF in all tumors was .72 ± .04 standard error of the mean (SEM) for FLAIR hyperintense and .68 ± .03 SEM for contrast-enhancing selleck kinase inhibitor regions, suggesting DSC had about a 3:1 higher dynamic range of CBF measurements

compared to ASL. Similarly for glioblastoma patients, a statistically significant linear was observed in FLAIR (Pearson’s correlation coefficient, R2 = .829, P < .0001) and contrast-enhancing regions (Pearson's correlation coefficient, R2 = .872, P < .0001). The linear relationship between ASL and DSC estimates of CBF in glioblastomas was similar to that of all tumors, measuring .74 ± .05 SEM for FLAIR and .66 ± .04 SEM for contrast-enhancing regions. These results suggest overall estimates of tumor blood flow may be similar between the two techniques, albeit to a different level of sensitivity and dynamic range. Surprisingly, only a minority of patients examined in the current study demonstrated a statistically significant linear correlation between DSC and ASL measurements of relative CBF on a voxel-wise basis for areas of FLAIR and contrast-enhanced regions. As illustrated in Figure

3A, some patients did illustrate a strong voxel-wise association Tenoxicam between the two measurements of CBF, specifically showing a 2:1 correspondence (slope ∼.5) between DSC and ASL DSC. The vast majority of patients, however, had voxel-wise relationships similar to those illustrated in Figure 3B, where no apparent linear relationship was evident. Approximately 31% of glioblastoma patients (4 of 13) demonstrated a significant voxel-wise linear correlation between DSC and ASL measurements of CBF with FLAIR hyperintense regions and only 38% of glioblastoma patients (5 of 13) showed a significant correlation in contrast-enhancing regions (Chi-Squared Goodness of Fit, χ2red > 1.0, P < .05). Interestingly, both patients with anaplastic astrocytoma (WHO III) had a significant voxel-wise correlation between DSC and ASL measurements of CBF in both FLAIR and postcontrast regions of interest (Chi-Squared Goodness of Fit, χ2red > 1.0, P < .05).