Table 2 Classification performance by the five DTI measures separately and all features together with three algorithms.

DTI index	Classifier	Balanced accuracy (%)	Recall (%)	Precision (%)	F1 (%)	ROC-AUC
FA volume	AdaBoost	55	78	69	71	0.60
	Gboost	47	66	63	62	0.49
	SVM	60	69	74	69	0.68
FA	AdaBoost	50	72	67	67	0.65
	Gboost	46	68	61	63	0.48
	SVM	56	81	72	74	0.52
Trace	AdaBoost	71	84	83	81	0.81
	Gboost	68	88	80	81	0.77
	SVM	57	72	73	70	0.71
AD	AdaBoost	48	72	66	67	0.48
	Gboost	69	88	80	82	0.71
	SVM	52	70	68	66	0.58
RD	AdaBoost	70	86	83	82	0.80
	Gboost	66	85	80	80	0.80
	SVM	57	67	72	67	0.69
All features	AdaBoost	67				0.78

ROC-AUC receiver operator characteristic curve-area under the curve, FA fractional anisotropy, AD axial diffusivity, RD radial diffusivity, AdaBoost Adaptive Boosting, GBoost Gradient Boosting, SVM Support Vector Machines.

Quick links

Search