Table 4:

Twenty highly correlated features with radiologist’s spiculation score in univariate analysis. T_s: the threshold to binarize the radiologist’s spiculation score

Rank	Feature name	AUC				Corr
		Ts > 1	Ts > 2	Ts > 3	Average	ρ
1	Ns	0.73	0.79	0.81	0.78	0.44
2	Roundness(mesh)	0.73	0.83	0.82	0.79	−0.44
3	SD LRE	0.74	0.80	0.76	0.77	−0.44
4	Mean ϵ	0.72	0.80	0.81	0.77	−0.41
5	Minimum ϵ	0.72	0.79	0.79	0.77	−0.41
6	Mean LRLGE	0.72	0.76	0.75	0.74	−0.40
7	SD LRLGE	0.71	0.77	0.77	0.75	−0.39
8	Weighted Principal Moment 1	0.71	0.76	0.78	0.75	0.39
9	Median ϵ	0.71	0.79	0.79	0.76	−0.39
10	Variance ϵ	0.70	0.78	0.79	0.76	0.39
11	Maximum ϵ	0.71	0.77	0.78	0.75	0.38
12	SD CS	0.71	0.73	0.71	0.72	−0.38
13	2D Weighted Principal Moment 1	0.71	0.74	0.76	0.73	0.37
14	s1	0.70	0.77	0.79	0.75	−0.37
15	SD LGRE	0.70	0.73	0.73	0.72	−0.36
16	2D Roundness(voxel)	0.68	0.78	0.77	0.74	−0.34
17	s2	0.68	0.75	0.77	0.73	0.34
18	SD SRLGE	0.68	0.72	0.71	0.70	−0.33
19	Mean Energy	0.69	0.69	0.66	0.68	−0.33
20	2D Sum	0.67	0.72	0.73	0.71	−0.32