Table 2.

Disease binary classification based on individual brain structures.

Case		Midbrain	Pons	Midbrain/pons	V3	Caudate	Putamen	Pallidum
Normal vs. PSP	V-Net	$0.73\pm 0.06$	${0.69\pm 0.03}^{\ast }$	$0.65\pm 0.08$	${0.83\pm 0.09}^{\ast }$	$0.54\pm 0.04$	${0.74\pm 0.01}^{\ast }$	${0.78\pm 0.05}^{\ast }$
	UNETR	$0.69\pm 0.06$	${0.64\pm 0.05}^{\ast }$	$0.60\pm 0.07$	${0.84\pm 0.08}^{\ast }$	$0.57\pm 0.03$	${0.69\pm 0.03}^{\ast }$	$0.76\pm 0.05$
	FS	$0.70\pm 0.06$	$\mathbf{0.89}\pm \mathbf{0.05}$	$0.65\pm 0.08$	$0.82\pm 0.02$	$0.56\pm 0.06$	$0.62\pm 0.09$	$0.72\pm 0.11$
Normal vs. MSA-P	V-Net	$0.63\pm 0.05$	$0.60\pm 0.05$	$\mathbf{0.73}\pm \mathbf{0.03}$	$\mathbf{0.73}\pm \mathbf{0.03}$	$0.61\pm 0.02$	$\mathbf{0.73}\pm \mathbf{0.03}$	$0.67\pm 0.10$
	UNETR	$0.61\pm 0.05$	$0.67\pm 0.06$	$0.70\pm 0.05$	$0.70\pm 0.03$	$0.59\pm 0.03$	$\mathbf{0.73}\pm \mathbf{0.03}$	$0.65\pm 0.09$
	FS	$0.64\pm 0.04$	$0.65\pm 0.04$	$0.70\pm 0.05$	$\mathbf{0.73}\pm \mathbf{0.03}$	$0.60\pm 0.06$	$0.70\pm 0.03$	$0.66\pm 0.08$
Normal vs. MSA-C	V-Net	$0.76\pm 0.11$	$0.90\pm 0.04$	$\mathbf{0.91}\pm \mathbf{0.02}$	$0.56\pm 0.02$	$0.61\pm 0.01$	$0.65\pm 0.11$	$0.66\pm 0.09$
	UNETR	$0.73\pm 0.08$	$0.86\pm 0.06$	$0.81\pm 0.15$	${0.58\pm 0.01}^{\ast }$	$0.57\pm 0.04$	$0.62\pm 0.10$	$0.66\pm 0.07$
	FS	$0.76\pm 0.10$	$0.90\pm 0.04$	$\mathbf{0.91}\pm \mathbf{0.02}$	$0.56\pm 0.01$	$0.62\pm 0.10$	$0.65\pm 0.10$	$0.71\pm 0.09$
Normal vs. PD	V-Net	$0.55\pm 0.02$	$0.53\pm 0.02$	$0.57\pm 0.04$	$0.61\pm 0.07$	$0.57\pm 0.02$	$0.55\pm 0.03$	$0.56\pm 0.02$
	UNETR	$0.58\pm 0.03$	$0.55\pm 0.03$	$0.53\pm 0.04$	$\mathbf{0.63}\pm \mathbf{0.06}$	$0.55\pm 0.04$	$0.54\pm 0.05$	$0.54\pm 0.03$
	FS	$0.56\pm 0.02$	$0.52\pm 0.01$	$0.57\pm 0.04$	$0.61\pm 0.08$	$0.54\pm 0.02$	$0.57\pm 0.01$	$0.54\pm 0.02$
PD vs. PSP	V-Net	$0.71\pm 0.07$	$0.67\pm 0.03$	$0.58\pm 0.07$	$0.74\pm 0.09$	$0.54\pm 0.02$	$0.67\pm 0.03$	$\mathbf{0.75}\pm \mathbf{0.03}$
	UNETR	$0.67\pm 0.09$	$0.63\pm 0.07$	$0.57\pm 0.08$	$0.72\pm 0.10$	$0.52\pm 0.01$	$0.63\pm 0.01$	$0.72\pm 0.05$
	FS	$0.69\pm 0.06$	$0.65\pm 0.04$	$0.58\pm 0.07$	$0.74\pm 0.11$	$0.52\pm 0.01$	$0.62\pm 0.03$	$0.71\pm 0.08$
PD vs. MSA-P	V-Net	$0.59\pm 0.07$	$0.67\pm 0.02$	$0.58\pm 0.06$	$0.65\pm 0.08$	$0.58\pm 0.02$	${0.67\pm 0.01}^{\ast }$	$0.65\pm 0.07$
	UNETR	$0.56\pm 0.02$	$0.64\pm 0.02$	${0.67\pm 0.04}^{\ast }$	$0.59\pm 0.05$	$0.61\pm 0.03$	${0.66\pm 0.01}^{\ast }$	$0.63\pm 0.05$
	FS	$0.59\pm 0.03$	$0.66\pm 0.02$	$0.58\pm 0.06$	$0.65\pm 0.08$	$0.59\pm 0.04$	$\mathbf{0.69}\pm \mathbf{0.04}$	$0.66\pm 0.07$
PD vs. MSA-C	V-Net	$0.74\pm 0.06$	$0.90\pm 0.03$	$\mathbf{0.94}\pm \mathbf{0.02}$	$0.56\pm 0.08$	$0.59\pm 0.06$	$0.50\pm 0.07$	$0.64\pm 0.06$
	UNETR	$0.69\pm 0.02$	$0.82\pm 0.11$	$0.81\pm 0.16$	$0.56\pm 0.08$	$0.58\pm 0.01$	$0.57\pm 0.05$	$0.62\pm 0.05$
	FS	$0.71\pm 0.06$	$0.90\pm 0.03$	$\mathbf{0.94}\pm \mathbf{0.02}$	$0.57\pm 0.08$	$0.59\pm 0.06$	$0.59\pm 0.08$	$0.69\pm 0.10$

Segmentation AUC of CNN-based V-Net, ViT-based UNETR, and FS. Mean ± standard deviation for threefold cross-validation and midbrain-to-pons ratio segmentation are listed.

*$p<0.05$ indicates a significant difference in AUC between the DL models and FS.

The best result for each volume segmentation method based on FS and DL in binary classification is shown in bold.