Cobb Angle Measurement of Spine from X-Ray Images Using Convolutional Neural Network

Table 3

Quantitative evaluation of the segmented results of U-Net, Residual U-Net, and Dense U-Net.

Methods	Dice similarity coefficient (DSC)	Jaccard similarity coefficient (JS)	Mean square error (MSE)	Accuracy (AC)	Sensitivity (SE)	Specificity (SP)
U-Net	0.941 ± 0.034	0.891 ± 0.057	0.030 ± 0.016	0.961 ± 0.022	0.980 ± 0.016	0.945 ± 0.038
Residual U-Net	0.951 ± 0.027	0.908 ± 0.046	0.025 ± 0.012	0.969 ± 0.016	0.982 ± 0.013	0.958 ± 0.029
Dense U-Net	0.948 ± 0.028	0.902 ± 0.048	0.027 ± 0.013	0.966 ± 0.017	0.982 ± 0.014	0.952 ± 0.031