Fully automated measurement on coronal alignment of lower limbs using deep convolutional neural networks on radiographic images

Table 2 The results of parameter calculation of lower limbs

		P value (95% CI)	ICC	Pearson correlation	MAE	RMSE
mLDFA ( °)	Right	0.212 (0.180, 0.225)	0.982	0.964	0.642	0.849
mLDFA ( °)	Left	0.194 (0.163, 0.202)	0.980	0.961	0.656	0.847
MPTA ( °)	Right	0.312 (0.193, 0.327)	0.988	0.976	0.638	0.872
MPTA ( °)	Left	0.332 (0.209, 0.344)	0.991	0.982	0.602	0.813
LDTA ( °)	Right	0.312 (0.193, 0.327)	0.939	0.887	1.986	2.451
LDTA ( °)	Left	0.394 (0.209, 0.344)	0.915	0.850	2.048	2.512
JLCA ( °)	Right	0.194 (0.158, 0.212)	0.949	0.907	0.825	1.092
JLCA ( °)	Left	0.224 (0.207, 0.261)	0.942	0.891	0.716	0.969
mechanical axis of lower limbs (mm)	Right	0.234 (0.207, 0.261)	1.000	1.000	1.124	1.416
mechanical axis of lower limbs (mm)	Left	0.194 (0.167, 0.237)	1.000	1.000	1.032	1.321

mLDFA, Mechanical lateral angle of distal femur MPTA, Medial angle of proximal tibia LDTA, Lateral angle of distal tibia JLCA, Joint convergent angle of articular surface ICC, Intraclass correlation coefficient MAE, Mean absolute error RMSE, Root mean squared error

ISSN: 1471-2474