Table 3.

The performance of different models using T1 and SWI sequence in binary classification tasks (PD vs MSA)

Sequence	Models	Average
		Sen	Spec	ACC	AUC
SWI	LR	0.750	0.811	0.785	0.864
	SVM	0.714	0.797	0.769	0.837
	LGBM	0.768	0.851	0.815	0.883
T1	LR	0.732	0.743	0.738	0.827
	SVM	0.696	0.824	0.769	0.853
	LGBM	0.714	0.811	0.769	0.844
SWI + T1	LR	0.821	0.757	0.785	0.848
	SVM	0.768	0.824	0.800	0.876
	LGBM	0.857	0.851	0.854	0.881

Sen sensitive, Spec specificity, ACC accuracy, AUC area under the curve, SWI susceptibility weighted imaging, T1 T1 weighted imaging, LR logistic regression, SVM support vector machine, LGBM light gradient boosting machine