Jul 2026
HCDU-net and the shape of precision: advancing liver tumor segmentation with statistical geometry and adaptive deep learning using U-net
The Hybrid Convolutional Deep U-Net is presented, an improved U-Net that incorporates multi-scale atrous convolutions, residual upsampling branches, and refined skip connections to capture contextual features and retain precise edge details across different tumor scales.
Chitra Jain, Puja Das, Ansul et al.
· Evolutionary Intelligence · 0 citations