Table 4 Test accuracies for different CNN architectures used on the RSNA dataset.

From: An efficient deep learning based approach for automated identification of cervical vertebrae fracture as a clinical support aid

CNN Model

Test accuracy (%)

DenseNet-121

97.73

Inception v3

97.25

Inception-ResNet-v2

97.83

Inception-ResNet-v2 U-Net

98.31

Inception-ResNet-v2 U-Net

(fine-tuned)

98.44