Cell Phenotype Classification Based on Joint of Texture Information and Multilayer Feature Extraction in DenseNet

Table 3

The performance evaluation of the proposed approach based on different classifiers and different DenseNet types in terms of accuracy (%) on the Hep-2 dataset.

DenseNet type	Classifier
	1-NN	3-NN	5-NN	MLP	Random forest	SVM
121	85.91	86.64	86.68	91.52	90.10	91.24
169	86.90	88.90	87.88	95.59	95.06	95.36
201	86.87	87.76	87.90	94.08	93.46	93.75
264	86.11	86.42	86.37	91.33	90.74	91.30