Clustered nuclei splitting based on recurrent distance transform in digital pathology images

EURASIP Journal on Image and Video Processing

Table 3 Performance of the proposed method in comparison to the state-of-the-art frameworks

	Methods	TPR	PPV	F1	MDR	FDR	USR	OSR	Jaccard	Time
Classical	None	0.647	0.766	0.686	0.03	0.15	0.33	0.08	0.72
	Watershed	0.723	0.754	0.727	0.04	0.14	0.25	0.09	0.72	155
	Proposed method	0.839	0.667	0.734	0.06	0.13	0.11	0.18	0.71	105
	H-minima	0.738	0.729	0.724	0.05	0.14	0.23	0.12	0.71	129
	Huang	0.676	0.513	0.576	0.31	0.25	0.02	0.19	0.52	77
	Kong	0.785	0.611	0.672	0.06	0.11	0.17	0.25	0.70	277
	Mouelhi	0.778	0.587	0.652	0.07	0.12	0.16	0.26	0.70	292
Frameworks	IJ_IHCtoolbox	0.541	0.848	0.643	0.43	0.03	0.05	0.12	0.63
	Qupath_def	0.796	0.510	0.612	0.16	0.25	0.05	0.20	0.66
	Qupath_exSep	0.869	0.576	0.682	0.08	0.21	0.06	0.18	0.74
	Tmarker	0.265	0.587	0.334	0.72	0.37	0.05	0.03	0.43
DL	Chen	0.776	0.700	0.724	0.20	0.25	0.04	0.04	0.55
	U-net	0.459	0.724	0.555	0.50	0.20	0.09	0.07	0.64