Deep Transfer Learning for Automatic Prediction of Hemorrhagic Stroke on CT Images

Table 1

Comparison of performance measures proposed transfer learning models.

Method	Accuracy	Sensitivity	Specificity	AUC	Training time (sec)
VGG-16	0.931	0.912	0.931	0.965	1849
GoogleNet (InceptionV3)	0.989	0.974	0.986	0.988	1809
ResNet-50	0.982	0.971	0.993	0.984	1835
Proposed model	0.996	0.994	0.997	1.000	1814