Medical image denoising method

A clinical medical image denoising method for clinical analysis and/or medical intervention, comprising:

• acquiring a clinical medical image by imaging a human body with an image acquisition system,
• receiving the clinical medical image on a computing device having a memory and a processor from the image acquisition system;
• and
• processing, with the processor of the computing device, the clinical medical image with a nonlinear diffusion coefficient based filter having a diffusion coefficient that is calculated using gradient vector orientation information in the clinical medical image to provide a denoised image,
• displaying the denoised image on a display device,
• wherein the nonlinear diffusion coefficient based filter is defined by following partial-differential-equation,
• ∂
• I
• ⁡
• (
• x
• ,
• y
• ,
• t
• )
• ∂
• t
• =
• ∇
• ·
• [
• c
• ⁡
• (
• x
• ,
• y
• ,
• t
• )
• ⁢
• ∇
• I
• ⁡
• (
• x
• ,
• y
• ,
• t
• )
• ]
• ,
• ⁢
• I
• ⁡
• (
• t
• =
• 0
• )
• =
• I
• 0
• ,
• where I(x,y,t) represents an image, (x, y) represents a position of pixel in the image, t is a parameter indicating a scale level in a scale space of the image I(x,y,t), I0 represents an initial image with t=0, ∇, and ∇ are divergence and gradient operators, respectively, and c (x, y, t) is the diffusion coefficient at a pixel located at position (x, y), and
• wherein the diffusion coefficient is defined as,
• c
• ⁡
• (
• x
• ,
• y
• ,
• ⁢
• t
• )
• =
• exp
• ⁡
• [
• -
• (
• 
• ∇
• I
• σ
• ⁡
• (
• x
• ,
• y
• ,
• t
• )
• 
• ×
• α
• σ
• ⁡
• (
• x
• ,
• y
• ,
• t
• )
• k
• )
• 2
• ]
• ,
• where Iσ(x,y,t) represents a smoothed image that is the initial image I0 smoothed with a Gaussian kernel having a width of σ, k is a noise threshold ασ(x, y, t), is a term carrying gradient vector orientation information.