Deep learning based Low-dose CT image denoising

DIAL Denoising

X-ray computed tomography (CT) is one of the most important imaging modalities in modern hospitals and clinics. However, there radiation poses a potential risk to the patient because x-rays could cause genetic damage and induce cancer with a probability that is proportionate to the radiation dose. Lowering the radiation dose increases the noise and artifacts in reconstructed images, which can compromise diagnostic information. Hence, extensive efforts have been made to design better image reconstruction or image processing methods for low-dose CT (LDCT). In this project, we develop a generative adversarial network with the Wasserstein distance (WGAN) as the discrepancy measure between distributions and a perceptual loss that computes the difference between images in an established feature space. WGAN is used to allow denoised CT images to share the same distribution as that of normal dose CT images. In particular, we treat the LDCT denoising problem as a transformation from LDCT to NDCT images. WGAN provides a good distance estimation between the LDCT and NDCT image distributions. Meanwhile, the perceptual loss can allow for the retention of the content of an image after denoising. The project is a result of collaborative efforts from multiple parties including Harvard Medical School and Prof. Ge Wang’s research group at RPI.

Source code can be found at