Skip to main content
. 2023 Jan 16;13:791. doi: 10.1038/s41598-023-27815-w

Table 5.

Detailed comparison of the proposed method with the existing models using three different datasets.

Methods Datasets Method description Evaluation
Dice IoU
Human8 JSRT Annotated by humans 0.9720 0.9460
JSRT 0.9740 0.9220
Ronneberge et al.22 MCCXR U-Net:encoder–decoder CNN 0.9547 0.9256
SCXR 0.9532 0.9119
Suza et al.23 MCCXR AlexNet for patch classifcation and ResNet18 based reconstruction 0.9400 0.8800
Novikov et al.16 JSRT FCN Inverted-Net 0.9685 0.9490
Lei et al.56 JSRT GAN based segmentation 0.9747 0.9508
DED-CNN JSRT Dual encoder–decoder structure 0.9760 0.9550
MCCXR 0.9767 0.9548
SCXR 0.9568 0.9184

The bolded values represent the best performances.