A deep convolutional neural network to predict the curve progression of adolescent idiopathic scoliosis: a pilot study

Table 3 Predictive values of curve progression with or without pre-trained deep convolutional neural network and fine-tuning

	AUC	Accuracy	Specificity	Sensitivity
Pre-trained DCNN
ROI1	0.53	0.55	0.57	0.53
ROI2	0.44	0.5	0.5	0.5
ROI3	0.7	0.69	0.77	0.61
Without transfer learning
ROI3	0.68	0.64	0.7	0.57

ISSN: 1471-2474