Figure 2.

Comparison of the bootstrapped CI for data smoothed with a 4, 8 or 12 mm FWHM isotropic Gaussian kernel. From A to D, histograms of the grey matter values across subjects (blue histograms) are plotted for the maximum of the average image, the standard deviation image, the kurtosis image and the skewness image obtained with the 8 mm FWHM smoothing kernel. The confidence interval (CIs) evolution graphics show the size of the CIs after each bootstrap. As illustrated, CI sizes differed depending on the smoothing kernel size. However, in all cases, CI sizes converged (i.e. were stable) after 3000 resamples (vertical black line). Similarly, the final histograms of the means obtained after 5000 resamples (red histograms titled 'Means distributions'), show that the estimated data ranges differed with the smoothing kernel size but that, in all cases, it was sufficient to obtain more closely normally distributed data.

Pernet et al. BMC Neuroscience 2009 10:67   doi:10.1186/1471-2202-10-67
Download authors' original image