Table 5 Computational cost analysis of ACU-Net and existing models.

From: Enhancing brain tumor segmentation using attention based convolutional UNet on MRI images

Model

Dataset

Params (M)

FLOPs (G)

Inference Time (s)

3D-UNet4

BRATS 2018

19.0

145.6

0.95

HTTU-Net5

BRATS 2018

21.5

158.4

1.05

RMU-Net27

BRATS 2018

17.8

139.2

0.87

CNN10

BRATS 2018

15.2

120.5

0.75

CNN11

BRATS 2018

16.9

125.8

0.80

3D-UNet12

BRATS 2018

20.3

148.7

1.00

ACU-Net (Proposed)

BRATS 2018

22.1

165.2

1.10

Back to article page