Skip to main content
Fig. 3 | BMC Cancer

Fig. 3

From: Convolutional neural network-based magnetic resonance image differentiation of filum terminale ependymomas from schwannomas

Fig. 3

The receiver-operating curves (ROC) for the image-level external test set for 3 CNN models. The area under the curve (AUC) can summarize the diagnostic effect of different models. The CNN using A EfficientNet-b2 model had an average AUC of 0.81. B ResNet-50 model had an average AUC of 0.72. C Inception-v3 had an average AUC of 0.84. The blue, green and orange dot lines represent the diagnostic efficacy of T1-WI, T2-WI and CE-T1 MRI modalities, respectively

Back to article page