Computer Aided Diagnostic Support System for Skin Cancer: A Review of Techniques and Algorithms

Table 3

Measures for evaluating performance of a classifier.

Evaluation parameters
Accuracy=TP+TNTP+TN+FP+FN×100%
Diagnosticaccuracy=TPTP+FP+FN
Sensitivity=TPTP+FN×100%
	Specificity=TNTN+FP×100%
ROC curve − a plot of the true positive TP-rate versus false positive FP-rate
Positivepredictivevalue=TPTP+FP×100%
Negativepredictivevalue=TNTN+FN×100%
Errorprobability=FP+FNTP+TN+FP+FN×100%
Index of suspicion =TP+FPTP+FN×100%
LR+=Sensitivity1−Specificity
LR−=1−SensitivitySpecificity
Diagnostic odds ratio [191], DOR =TP/FNFP/TN
Distance of a real classifier from the ideal one 𝑑class=(1−Se)^2+1−Sp^2