Accurate mandibular canal segmentation of dental CBCT using a two-stage 3D-UNet based segmentation framework

BMC Oral Health

Table 1 Results of the two-stage 3D-UNet segmentation model

Model	Data grouping	DSC	95% HD (mm)
Coarse segmentation	Train dataset	0.847 ± 0.038	0.508 ± 0.564
	Validation dataset	0.860 ± 0.040	0.630 ± 0.585
	Test dataset	0.827 ± 0.057	1.027 ± 1.422
Fine segmentation	Train dataset	0.872 ± 0.028	0.417 ± 0.170
	Validation dataset	0.884 ± 0.023	0.352 ± 0.104
	Test dataset	0.875 ± 0.045	0.442 ± 0.379

ISSN: 1472-6831