Simple noise reduction for diffusion weighted images

Radiol Phys Technol. 2016 Jul;9(2):221-6. doi: 10.1007/s12194-016-0350-9. Epub 2016 Mar 16.

Abstract

Our purpose in this study was to reduce the noise in order to improve the SNR of Dw images with high b-value by using two correction schemes. This study was performed with use of phantoms made from water and sucrose at different concentrations, which were 10, 30, and 50 weight percent (wt%). In noise reduction for Dw imaging of the phantoms, we compared two correction schemes that are based on the Rician distribution and the Gaussian distribution. The highest error values for each concentration with use of the Rician distribution scheme were 7.3 % for 10 wt%, 2.4 % for 30 wt%, and 0.1 % for 50 wt%. The highest error values for each concentration with use of the Gaussian distribution scheme were 20.3 % for 10 wt%, 11.6 % for 30 wt%, and 3.4 % for 50 wt%. In Dw imaging, the noise reduction makes it possible to apply the correction scheme of Rician distribution.

Keywords: Correction scheme; Diffusion weighted imaging; Gaussian distribution; Magnetic resonance imaging (MRI); Probability distribution function (PDF); Rician distribution.

MeSH terms

  • Diffusion Magnetic Resonance Imaging*
  • Image Enhancement / methods*
  • Phantoms, Imaging
  • Signal-To-Noise Ratio*