Table 3.

Class-specific Accuracies for Each Classifier Family

	Healthy	Emphysema	Ground glass	Fibrosis	Micronodules	Macronodules
Naive Bayes	0.9161	0.8753	0.7873	0.7454	0.3778	0.3076
k-NN	0.9104	0.9978	0.7369	0.8926	0.91	0.3605
J48	0.7568	0.9419	0.6803	0.8821	0.7555	0.3054
MLP	0.7290	0.9707	0.6756	0.8751	0.8035	0.2461
SVM	0.9242	0.9874	0.7731	0.9218	0.8907	0.4250

Best performances are highlighted in bold. SVM reached three times the best accuracy and is no more than 2% behind the best performance over all classes