基于KAN与CKAN优化的医学图像分割模型
|
|
娄世猛,邵玉斌,杜庆治,唐菁敏,张赜涛
|
Medical image segmentation model based on KAN and CKAN optimization
|
|
Shimeng LOU,Yubin SHAO,Qingzhi DU,Jingmin TANG,Zetao ZHANG
|
|
| 表 4 UNet、KUNet、nnUNet、Swin-UNet模型在各数据集上的计算效率 |
| Tab.4 Calculation efficiency of UNet, KUNet, nnUNet and Swin-UNet model on each dataset |
|
| 数据集 | $ {E}_{\text{UNet}} $ | $ {E}_{\text{nnUNet}} $ | $ {E}_{\text{Swin-UNet}} $ | $ {E}_{\text{KUNet}} $ | $ {T}_{\text{UNet}} $/s | $ {T}_{\text{nnUNet}} $/s | $ {T}_{\text{Swin-UNet}} $/s | $ {T}_{\text{KUNet}} $/s | | LiTS | 100 | 100 | 100 | 100 | 557.11 | 4344 | 544 | 998.17 | | CORN | 2000 | 100 | 2000 | 2000 | 37674.04 | 5507 | 15380 | 57941.40 | | DRIVE | 5000 | 5000 | 5000 | 10000 | 1630.20 | 63932 | 1659 | 5458.41 | | Lungs | 100 | 100 | 100 | 50 | 262.87 | 5509 | 383 | 241.84 |
|
|
|