Skip to main content
. 2014 Jun;35(6):1219–1225. doi: 10.3174/ajnr.A3830

Fig 7.

Fig 7.

Impact of data quality on variability of tract parameters. Diffusion-weighted images with high outlier percentages (>10%) resulted in a significantly increased variability of mean fractional anisotropy values compared with data with fewer data outliers (<10%) if linear least squares was used (Levene test for equality of variances, P < .05). With application of robust estimation of tensors by outlier rejection, there was no difference in variability with regard to data quality.