In the original article, there was a mistake in Table 3 as published. The labels for [Hb] and OEF0 were swapped. The corrected Table 3 appears below.
TABLE 3.
Results of a bivariate regression of OEF0 against CBF0 and [Hb] for 30 healthy volunteers analyzed with the ML (ensemble of MLPs) and rNLS fitting methods.
| Predictor | ML β1 (p value) | rNLS β1 (p value) |
|---|---|---|
| [Hb] | −1.42 (0.001) | −2.23 (0.001) |
| CBF | −0.07 (0.44) | −0.37 (0.005) |
| Intercept | 61.95 (<0.001) | 89.48 (<0.001) |
In the original article, there was an error. Reference to Table 3 in the text had the labels for [Hb] and OEF0 swapped. A correction has been made to Results, In-vivo, paragraph 5:
“Table 3 reports the results of a bivariate regression of OEF against [Hb] and CBF for both analysis methods. The slopes of the relationship between OEF and [Hb] are similar to that reported in healthy subjects by Ibaraki et al. (2010), −1.75 Hb (g/dL). As per Ibaraki et al. the relationship between CBF and OEF did not reach significance (p = 0.44) for the ML approach, however a significant negative correlation was observed in the rNLS analysis (p = 0.005). A univariate analysis of CMRO2,0 against CBF0 is consistent with that observed in healthy controls by Powers et al. (2011) (β1 = 0.2) for both analysis methods, β1 = 0.32 (p < 0.001) and β1 = 0.24 (p < 0.001) for the ML and rNLS approaches respectively.”
The authors apologize for these errors and state that this does not change the scientific conclusions of the article in any way. The original article has been updated.