Table 2 Comparison of various lower encoders (for clarity, these encoders are divided into two types for comparison).

From: Multi-view hybrid encoder U-Net for 3D renal vascular medical image segmentation

Encoder Arch.

Lower encoder

Loss

FPR

Recall

DSC

MSD

CNNs + CNNs

Efficientnet_b331

0.4303

0.135

0.924

0.897

0.808

Ese Vovnet39b32

0.4337

0.092

0.938

0.924

0.832

Seresnext50_32 × 4d33

0.4348

0.140

0.934

0.900

0.811

Convnext Small34

0.4187

0.076

0.950

0.938

0.846

DM Nfnet F035

0.4195

0.076

0.946

0.936

0.842

Tiny ViT 21 m36

0.4254

0.182

0.961

0.897

0.801

FastVit Sa1237

0.4279

0.082

0.916

0.917

0.825

CNNs + ViTNets

Sam2 Hiera Small38

0.4295

0.087

0.956

0.936

0.839

Swin Tiny39

0.4221

0.080

0.948

0.935

0.842

NextViT Small40

0.4247

0.114

0.940

0.915

0.826

 

Proposed (CoaT Lite Small)41

0.4193

0.063

0.941

0.939

0.852

  1. Significant values are in bold.
Back to article page