Influence of growth structures and fixed appliances on automated cephalometric landmark recognition with a customized convolutional neural network

Table 6 Overall model performance for all observed patient groups

	Mean X -Error	Mean Y -Error	Mean Radial Error	Standard Deviation	SDR (Successful detection rates) %
	(mm)	(mm)	MRE (mm)	SD (mm)	SDR % *2 mm*	SDR % *2.5 mm*	SDR % *3 mm*	SDR % *4 mm*
A-Point	0.74	0.90	1.29	0.77	89.57	93.91	96.09	99.13
Ap-Superior	1.08	1.38	1.93	1.30	66.74	75.43	84.57	93.48
Ap-Inferior	1.07	1.40	1.91	1.34	67.17	75.00	83.48	93.70
ANS	1.21	0.96	1.75	1.51	80.65	81.96	87.39	94.57
Art	0.79	0.79	1.25	0.85	89.57	92.39	95.87	98.48
B-Point	0.55	1.04	1.26	0.81	94.13	95.65	96.52	99.35
Ba	0.90	1.18	1.66	1.21	86.09	86.74	90.43	94.78
Me	1.00	0.75	1.38	0.93	85.87	89.57	93.91	97.39
Is-Superior	0.59	0.68	1.00	0.80	93.91	96.96	97.83	98.91
Is-Inferior	0.67	0.84	1.19	0.93	89.57	94.57	96.30	99.35
N	0.67	0.86	1.21	0.77	88.70	94.57	98.04	99.57
Pog	0.65	1.39	1.66	1.22	79.35	82.17	88.48	95.00
PNS	1.31	0.83	1.71	1.34	82.39	83.48	85.65	93.04
S	0.63	0.59	0.97	0.84	94.13	97.17	98.04	99.57
T1	1.00	1.02	1.61	1.21	83.26	85.00	89.35	94.57
T2	1.28	0.91	1.72	1.18	84.57	86.74	89.13	94.35
AVERAGE			1.47	1.06	84.73	88.21	91.94	96.58

ISSN: 1472-6831