A Novel Medical Image Denoising Method Based on Conditional Generative Adversarial Network

Algorithm 1

The training process of the proposed medical image denoising method based on conditional generative adversarial network.

1.Require: Set hyper-parameters: , batch size =1, ,

2.Get by adding some artificial noise in raw image

3.Obtain the corresponding gradient map by calculating the gradient for each pixel in

4.The median T of the gradient map is set as the threshold in , then obtain

5.Get the gradient enhancement images by adding up the and

6.Initialize the parameters of generator and discriminator

7.fordo

8.Sample a batch of raw image patches and the image to be processed patches

9.

10. Concatenate

11. Update the discriminator D by Adam optimizer according to the original GAN loss

12. Update the generator G by Adam optimizer according to the Equation (12)

13.end for