Skip to main content
. 2012 May 19;34(10):2635–2654. doi: 10.1002/hbm.22092

Table 11.

Results for model‐based and MAGeT Brain segmentations using manually generated template library using atlases given as mean (confidence interval)

Structure 3 Template Multi‐Atlas 3 Template MAGeT Multi‐Atlas
ANIMAL ANTs ANIMAL ANTs ANIMAL ANTs
Kappa
Striatum 0.885 (0.880–0.889) 0.888 (0.882–0.891) 0.891a (0.887–0.896) 0.894a (0.890–0.898) 0.892 (0.887–0.895) 0.900b (0.896–0.904)
Globus pallidus 0.790 (0.772–0.808) 0.797 (0.781–0.812) 0.815a (0.800–0.827) 0.805 (0.789–0.820) 0.811 (0.797–0.824) 0.819 (0.804–0.834)
Thalamus 0.883 (0.874–0.891) 0.887 (0.878–0.895) 0.895 (0.883–0.901) 0.895 (0.884–0.902) 0.891 (0.882–0.900) 0.902 (0.893–0.911)
Jaccard
Striatum 0.784 (0.774–0.793) 0.795 (0.790–0.803) 0.810a (0.797–0.812) 0.806a (0.799–0.813) 0.803 (0.797–0.811) 0.820b (0.814–0.826)
Globus pallidus 0.645 (0.622–0.670) 0.662 (0.643–0.685) 0.687a (0.667–0.707) 0.676 (0.654–0.697) 0.684 (0.664–0.703) 0.696 (0.675–0.717)
Thalamus 0.794 (0.781–0.811) 0.794 (0.784–0.811) 0.806 (0.792–0.822) 0.807 (0.792–0.821) 0.804 (0.790–0.819) 0.821 (0.809–0.837)

Results are given for multi‐atlas and MAGeT Brain segmentation with 3 templates (3 Template Multi‐Atlas and 3 Template MAGeT, respectively) and full multi‐atlas segmentation.

a

Represents significance levels of P < 0.05 for 3 Template MAGeT Brain as compared to its 3 Template Multi‐Atlas counterpart.

b

Represents significance levels of P < 0.05 for Multi‐Atlas segmentation compared to 3 Template Multi‐Atlas counterpart.

Only segmentations using the same nonlinear registration algorithms were compared to one another (e.g., only the model‐based segmentation using ANIMAL was compared to the MAGeT Brain segmentation using ANIMAL).