Table 5.

Average Dice coefficients (DCs) (±standard deviations) for automatic segmentation of myelin-like signals in deep brain region of eight test subjects with respect to two sets of manual annotations compared to the intra-rater reliability.

Method	Manual set 1		Manual set 2		Intra-raterreliability
	Average DC	p-value	Average DC	p-value
GMM	0.785 ± 0.085	0.257	0.754 ± 0.054	0.023	0.834 ± 0.043
GMM-PV	0.662 ± 0.210	0.060	0.626 ± 0.217	0.028
GMM-PV-MRF	0.843 ± 0.051	0.624	0.816 ± 0.046	0.413

The p-values indicate the outcomes of two-tailed Student’s t-tests with the intra-rater DCs at the 5% significance level. The performance of GMM-PV-MRF is close to the intra-rater reliability.