Estimation of Abnormal Cell Growth and MCG-Based Discriminative Feature Analysis of Histopathological Breast Images

Table 1

Comparative performance analysis of classifiers using the BreakHis dataset after and before graph minimization.

Classifier	Category	F-measure	Precision	Recall	Specificity	Sensitivity	Accuracy (%)
SVM	CG	0.4369	0.4737	0.4054	0.5495	0.4054	47.75
	MCG	0.8829	0.8512	0.9279	0.8378	0.9279	88.29
SVM-polynomial	CG	0.6744	0.9508	0.5225	0.9730	0.5225	74.77
	MCG	0.9683	0.9727	0.9640	0.9730	0.9640	96.85
SVM-Gaussian	CG	0.6780	0.9091	0.5404	0.9459	0.5404	74.32
	MCG	0.9636	0.9725	0.9550	0.9730	0.9550	96.40
SVM-RBF	CG	0.6845	0.8421	0.5766	0.8919	0.5766	73.42
	MCG	0.9541	0.9720	0.9369	0.9730	0.9369	95.50
Decision tree	CG	0.7064	0.7196	0.6937	0.7297	0.6937	71.17
	MCG	0.9633	0.9813	0.9459	0.9820	0.9459	96.40
Random forest	CG	0.7240	0.7273	0.7207	0.7297	0.7207	72.52
	MCG	0.9593	0.9636	0.9550	0.9640	0.9550	95.95
Nearest neighbor	CG	0.7163	0.7404	0.6937	0.7568	0.6937	72.52
	MCG	0.9464	0.9381	0.9550	0.9369	0.9550	94.59
LDA	CG	0.6569	0.7204	0.6036	0.7658	0.6036	68.47
	MCG	0.8018	0.8208	0.7838	0.8288	0.7838	80.63
Naive Bayes fitrensemble	CG	0.6915	0.8442	0.5856	0.8919	0.5856	73.87
	MCG	0.9493	0.9717	0.9279	0.9730	0.9279	95.05