Segmentation of white blood cells and comparison of cell morphology by linear and naïve Bayes classifiers

Table 3 Performance of segmentation

Type	The digital microscope collected image (dataset 1)						CellaVision dataset (dataset 2)
Type	Baso	Eos	Lym	Mono	Neu	Average	Baso	Eos	Lym	Mono	Neu	Average
Nucleus segmentation
Dice similarity	0.950	0.868	0.977	0.940	0.948	0.937	0.881	0.923	0.972	0.875	0.995	0.929
FPR	0.002	0.163	0.002	0.000	0.002	0.034	0.205	0.131	0.056	0.025	0	0.083
FNR	0.094	0.098	0.042	0.108	0.094	0.087	0.051	0.037	0.005	0.193	0.009	0.059
Cell segmentation
Dice similarity	0.955	0.836	0.919	0.825	0.899	0.887	0.913	0.947	0.963	0.945	0.966	0.947
FPR	0.008	0.275	0.072	0.225	0.089	0.134	0.002	0.092	0.052	0.061	0.038	0.049
FNR	0.078	0.100	0.089	0.146	0.112	0.105	0.159	0.023	0.023	0.052	0.030	0.057

ISSN: 1475-925X