The denoising on ct/x-ray images is still a challenge in

The denoising on CT/X-Ray images is still a challenge in
medical image processing, especially, on the photon-limited images. The state-of-the-art methods cannot solve
simultaneously the following tasks: high accuracy on both
photon-limited and photon-unlimited images, avoid adding artificial artifacts and the performance. The goal to develop an effective universal method that reduces multiple
types of noise is even more difficult challenge.
In this paper, we reviewed on the following methods:
MROF, ATV, ANLTV, HNIPM, PRBF, PURE-LET and
MS-VST. The PRBF is excellent choice if the execution
time is the most important. However, if the accuracy is
priority, non-local methods are recommended. If we need
to process the photon-limited images, the ANLTV, MSVST and adaptive VST-based methods are very good
choices. If we want to exploit the existing Gaussian denoising methods, we can use adaptive VST-based methods, including MS-VST.
During denoising process is performed, it is necessary
to avoid adding artificial structures, and one can choose
ATV or ANLTV methods that provide good denoising
performance without introducing discernible artifacts. In
this case, the VST-based methods can be used if they are
combined to the image structure preservation Gaussian denoising methods, such as BM3D [31], SAFIR [32] etc.
By the research trend, the VST-based approach is a
novel option by the criteria to create an “universal”
method to remove multiple type of noises. This approach
has potential if it is possible to expand the VST-based approach to apply to other signal dependent noises