Application of different wavelet families and thresholding methods for low-dose computed tomography denoising

X-ray Computed Tomography (CT) is one the predominant imaging devices. X-ray photons are generated in a vacuum tube, and directed towards the patient body. A reconstruction algorithm such as filtered back projection (FBP) is normally used to reconstruct the 3D-volume image of the patient from the projection data. Since CT imaging is based on transmission and absorption of X-ray photons, it will deposit radiation dose inside the patient, which has the risk of producing unhealthy biological events. However, reducing the radiation flux will generate quantum mottle noise in the CT images. In this study, wavelet denoising was successfully applied in order to reduce the noise in a low-dose CT image. The low-dose CT image had 75% less dose compared to its normal dose version. Different types of wavelets and thresholding methods were applied. The denoising performance of each one was evaluated using structural similarity index (SSIM). The results showed that the selection of thresholding method and threshold value are the most important factors in CT image denoising using wavelet transform.