A New Classification Method in Ultrasound Images of Benign and Malignant Thyroid Nodules Based on Transfer Learning and Deep Convolutional Neural Network
Table 4
Comparison of evaluation index obtained by different methods.
Method
F1 score
Precision
Recall
Accuracy
Twice transfer learning based on LeNet5
0.9748
0.9678
0.9671
0.9649
Once transfer learning based on LeNet5
0.9584
0.9788
0.9685
0.9635
Twice transfer learning based on VGG16
0.9657
0.9718
0.9781
0.9742
Once transfer learning based on VGG16
0.9604
0.9548
0.9685
0.9692
Twice transfer learning based on inception V3 (our method)
0.9874
0.9842
0.9684
0.9885
Once transfer learning based on inception V3 (our method)
