Table 2.

AUROC achieved by the compared tissue-specific CT image sub-segmentation methods on the HGSOC (POD and omental lesions) and RCC datasets.

Method	HGSOC (POD)		HGSOC (omentum)		RCC
	Hyper-dense	Hypo-dense	Hyper-dense	Hypo-dense	Hyper-dense	Hypo-dense
Fixed thresholding	0.781	0.534	0.674	0.573	0.667	0.771
Two-stage Otsu	0.682	0.689	0.623	0.716	0.500	0.802
Silhouette ＋ $K$-means	0.694	0.578	0.572	0.589	0.466	0.739
CH ＋ $K$-means	0.471	0.501	0.499	0.492	0.333	0.500
DB ＋ $K$-means	0.481	0.507	0.528	0.535	0.282	0.553
Gap ＋ $K$-means	0.628	0.754	0.519	0.599	0.490	0.719
sFCM${}_{1,0}$	0.901	0.987	0.937	0.987	0.943	0.981
sFCM${}_{1,1}$	0.901	0.987	0.937	0.987	0.943	0.981
sFCM${}_{1,2}$	0.901	0.987	0.937	0.987	0.943	0.981
ksFCM${}_{1,0}$	0.917	0.982	0.937	0.981	0.943	0.981
ksFCM${}_{1,1}$	0.901	0.982	0.937	0.981	0.943	0.981
ksFCM${}_{1,2}$	0.901	0.987	0.937	0.987	0.943	0.981