Table 7.

Summary of studies with the best performance.

	Name	Sample	Classification	Measures	ROIs	Acc	SS	SP	AUC
bvFTD vs Controls	Raamana et al., 2014	30 bvFTD	SVM	Surface displacements	L lateral ventricle		100	88	0.938
		14 C	Train/test
bvFTD vs AD	Canu et al., 2017	27 bvFTD	Random forest	Cortical thickness	Best 5 (L inferior parietal, R temporal pole, L isthmus cingulate, R inferior parietal, R precuneus)	82	80	87
		62 AD
FTD vs Controls	Davatzikos et al., 2008	12 FTD	SVM	RAVENS-GM and WM volume	PCA	100
		12 C	LOOCV
FTD vs AD	McMillan et al., 2014	72 FTD	Linear regression	Combination (Cortical thickness & DTI-FA)	Data-driven		89	89	0.874
		21 AD
			Train/test
FTD vs AD & Controls	Kuceyeski et al., 2012	18 FTD	Linear discriminant analysis	DWI-RD	Whole-brain parcellation	89.09	97.30	72.22
		18 AD
			LOOCV
		19 C
FTD vs other dementias	Vemuri et al., 2011	7 FTD	Differential-STAND	GM density	Whole-brain		84.4	93.8
			LOOCV
		48 AD
		20 DLB
		21 C4
nfvPPA vs Controls	Bisenius et al., 2017	6 nfvPPA	SVM	VBM-GM density	Whole-brain	91	88	94	0.94
		20 C	LOOCV
lvPPA vs Controls	Wilson et al., 2009	16 lvPPA	SVM	GM volume	PCA	100	100	100	1
		115 C	2-level CV
svPPA vs Controls	Bisenius et al., 2017	17 svPPA	SVM	VBM-GM density	ROI (a priori from meta-analyses)	100	100	100	1
		20 C	LOOCV
	Wilson et al., 2009	38 svPPA	SVM	GM volume	PCA	100	100	100	1
		115 C	2-level CV
svPPA vs nfvPPA	Wilson et al., 2009	32 nfvPPA	SVM	GM volume	PCA	89.1	84.4	93.8	0.964
		38 svPPA	2-level CV
lvPPA vs svPPA	Bisenius et al., 2017	11 lvPPA	SVM	VBM-GM density	Whole-brain	95	100	91	0.93
		17 svPPA	LOOCV
lvPPA vs nfvPPA	Wilson et al., 2009	32 nfvPPA	SVM	GM volume	PCA	81.3	81.3	81.3	0.879
		16 lvPPA	2-level CV