An Edge-Assisted Computing and Mask Attention Based Network for Lung Region Segmentation

Table 5

Results of EAM-Net and the state-of-the-art lung region segmentation methods on the Shenzhen dataset.

Methods	PA (%)	Dice (%)	JA (%)
Feature selection with BN [10]	81.22	67.85	51.09
Feature selection with MLP [10]	87.82	77.85	64.74
Feature selection with RF [10]	89.19	80.25	68.15
Feature selection and vote [10]	91.06	83.65	73.04
Bayesian feature pyramid network [21]	96.17	93.04	87.00
X-ray-Net [23]	97.21	94.92	90.33
Kim and Lee [32]	97.49	95.45	91.26
EAM-Net	98.20	96.13	92.27

Bold values represent the the highest performance for each performance metric.