Table 3.

Signed mean ​± ​standard deviation results for healthy knee participants comparing U-Net and U-Net++ two-dimensional (2D) deep-learning segmentations on transverse ultrasound images to subsequent three-dimensional (3D) reconstructed segmentations. The p-value corresponds to a comparison between 2D and 3D segmentation performance.

Method	Segmentation	sA/VPD (%)	sMSD (mm)	sHD (mm)
U-Net	2D Transverse	10.1 ​± ​4.71	0.10 ​± ​0.04	1.06 ​± ​0.46
	3D Reconstruction	−3.01 ​± ​13.1	0.19 ​± ​0.07	0.59 ​± ​1.73
	P-value	0.8371	0.0566	0.5829
U-Net++	2D Transverse	3.92 ​± ​7.79	0.08 ​± ​0.13	0.56 ​± ​1.18
	3D Reconstruction	28.2 ​± ​8.09	0.27 ​± ​0.07	1.56 ​± ​0.43
	P-value	0.021	0.002	0.007