Fig. 3

The ResNet-U-Net architecture and parameters utilized in this study are detailed. The architecture integrates the robust feature extraction capabilities of the Residual Network (ResNet) with the U-Net’s efficient segmentation framework. The ResNet serves as the backbone, employing residual blocks that allow the training of deeper networks by mitigating the vanishing gradient problem. The U-Net structure, with its characteristic skip connections, ensures the preservation of spatial details essential for accurate segmentation.