Soft Tissue/Bone Decomposition of Conventional Chest Radiographs Using Nonparametric Image Priors

Table 1

The average performance of MAP decomposition and locally weighted regression on the validation set with 100 standard chest radiographs. bsr denotes the bone suppression ratio. rmse-S and ssim-S denote the root mean squared error (rmse) and the structural similarity image measure (ssim) for the reconstructed soft tissue, respectively.

Method	bsr	rmse-S	ssim-S
MAP decomposition	0.715	0.414	0.927
Weighted regression	0.704	0.441	0.846