TABLE 7.

Component analysis of MixACB by two‐fold cross validation

	CSF		GM		WM		AVG
	DICE	MHD	DICE	MHD	DICE	MHD	DICE	MHD
CONV_3	0.951 ± 0.008	0.261 ± 0.021	0.922 ± 0.007	0.605 ± 0.046	0.907 ± 0.015	0.452 ± 0.041	0.927 ± 0.007	0.439 ± 0.032
AC_3	0.952 ± 0.010	0.261 ± 0.024	0.922 ± 0.008	0.604 ± 0.045	0.906 ± 0.014	0.453 ± 0.041	0.927 ± 0.008	0.440 ± 0.033
CONV_5	0.947 ± 0.012	0.276 ± 0.023	0.918 ± 0.008	0.619 ± 0.047	0.903 ± 0.016	0.463 ± 0.043	0.922 ± 0.008	0.453 ± 0.034
AC_5	0.952 ± 0.008	0.261 ± 0.022	0.920 ± 0.008	0.610 ± 0.046	0.904 ± 0.016	0.460 ± 0.043	0.925 ± 0.008	0.443 ± 0.033
MixACB	0.953 ± 0.008	0.254 ± 0.022	0.923 ± 0.007	0.601 ± 0.047	0.907 ± 0.015	0.452 ± 0.043	0.928 ± 0.008	0.436 ± 0.034

Note: The best values are highlighted in bold font. “CONV_3” denotes that the 3D convolution with a kernel size of 3; “AC_3” denotes that the 3D‐AC with a kernel size of 3; “CONV_5” denotes that the 3D convolution with a kernel size of 5; “AC_5” denotes that the 3D‐AC with a kernel size of 5.