Your privacy, your choice

We use essential cookies to make sure the site can function. We also use optional cookies for advertising, personalisation of content, usage analysis, and social media.

By accepting optional cookies, you consent to the processing of your personal data - including transfers to third parties. Some third parties are outside of the European Economic Area, with varying standards of data protection.

See our privacy policy for more information on the use of your personal data.

for further information and to change your choices.
Skip to main content

Table 2 Comparison of model complexity and computational performance

From: Diffusion-CSPAM U-Net: A U-Net model integrated hybrid attention mechanism and diffusion model for segmentation of computed tomography images of brain metastases

Model

Params (M)

FLOPs (G)

Total Training Time (h)

Inference Time/Per Slice (ms)

U-Net

31.75 M

31.29G

104.63 h

20 ms

U-Net + CAM

31.81 M

47.67G

105.52 h

27 ms

U-Net + CAM + SAM

31.94 M

42.83G

107.80 h

29 ms

U-Net + CAM + SAM + PAM (CSPAM-U-Net)

32.30 M

58.05G

110.17 h

35 ms

  1. Note: The total training time included the diffusion model. The inference time is the time of inference per slice on the validation set for the completed trained model
  2. CAM, Channel Attention Module; SAM, Spatial Attention Module; PAM, Positional Attention Module; Params, Parameters; GFLOPs, Giga Floating-point Operations Per Second
Back to article page

Radiation Oncology

ISSN: 1748-717X

Contact us