Application of deep learning and feature selection technique on external root resorption identification on CBCT images

Table 4 Comparison of classification performance between the deep learning-based systems

Models	Accuracy (Overall)	Accuracy (Weighted)	F1-score	Precision	Specificity	Error rate	AUC
RF + VGG	0.7882	0.82	0.7882	0.7882	0.7882	0.2117	0.95
FS + RF + VGG	0.8188	0.83	0.8188	0.8188	0.8188	0.18117	0.96
RF + EFNET	0.5529	0.61	0.5529	0.5529	0.5529	0.4471	0.84
FS + RF + EFNET	0.5906	0.61	0.5906	0.5906	0.5906	0.4094	0.85
SVM + VGG	0.7718	0.81	0.7718	0.7718	0.7718	0.2282	0.81
FS + SVM + VGG	0.7906	0.81	0.7906	0.7906	0.7906	0.2094	0.84
SVM + EFNET	0.7200	0.75	0.7200	0.7200	0.7200	0.2800	0.85
FS + SVM + EFNET	0.7671	0.79	0.7671	0.7671	0.7671	0.2329	0.85

ISSN: 1472-6831