Automatic estimation of noise to signal powerspectrum ratio in wiener filtering technique for computed tomography images
One of the most common medical imaging modalities used in examining internal structure of human body nowadays is Computed Tomography (CT). The main concept of CT imaging is to produce cross sectional images of interior anatomical structure inside the body. Furthermore CT image is very useful in dete...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/47561/1/FK%202013%2033R.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | One of the most common medical imaging modalities used in examining internal structure of human body nowadays is Computed Tomography (CT). The main concept of CT imaging is to produce cross sectional images of interior anatomical structure inside the body. Furthermore CT image is very useful in determining the extent of bone destruction and it is also an essential tool in evaluating the severity of diseases especially around specific location of the body. In various applications particularly medical imaging, degradation process normally occurred due to noise and degradation phenomena. The degradation phenomena, also known as Point Spread Function (PSF) may be caused by several factors, for instance, from out-of-focus blur, 2D Gaussian Blur, turbulence blur, motion blur and electronic noises.
In practice, image restoration is a process that tries to recover or reconstruct a degraded image by applying knowledge that causes the degradation phenomenon.
Prior knowledge of the degradation phenomenon and inverse filter must be obtained or estimated in order to restore back the image. Therefore restoration method is
concerned toward mathematically modelling the degradation PSF and applying the inverse process to recover the undegraded image.
In this research, the best possible restoration technique for CT image is presented. The whole research is concerned about Wiener filter and its parameters, which are power
spectrum of noise and signal, PSF and noise to signal ratio (NSR). The main objective of this study is to offer a method which is able to approximate a desirable value of
noise to signal power spectrum ratio (NSR) via frequency domain where it will be applied into the Wiener filter equation. Another parameter of Wiener filter that has
been considerate is PSF in which three typical types of PSF are compared. Furthermore an analytical comparison had been done regarding the competence of specific PSF in restoring these particular CT images. All those procedures utilized the information of the CT image itself. Quality of CT images are measured by using two types of mathematical evaluation assessments, which are Blur Metric and Blind Image Quality Assessment through Anisotropy (BIQAA). Therefore, a standard value of good quality image can be define in terms of blur and contrast.
Ultimately the proposed method provided an automatic estimation for NSR value compared to the previously or commonly technique. The previously or typically
method in obtaining the value of NSR usually by treating the value as constant and further proceed with utilize try and error method. The iteration process will halt when
it achieved the conditions where the CT image was assumed as better in quality and visualization. The process was manually done, inconvenient and hassle.
The proposed technique automatically provide the NSR value without required any external information. The novelty of the research is the proposed method will utilize
the information gathered from the image itself and automatically generate the desirable value of NSR. The proposed method was tested with sixty random CT images and all the images provide significant enhanced image in term of quality of visualization.
Generally, the method proposed yields a promising result on restoring CT image. However, the result is still depending on the complexity of the CT image with various organ structure, whether it consists of either complex or plain structure anatomy. The formed result of the image restoration will be less significantly visible if less
anatomy structures emerged out of the CT image. Finally, based on values presented by both quality assessments, it can be concluded that the proposed restoration technique
manage to yield CT image with better quality of visualization compared to the original CT image yielded by the CT scan machine. Furthermore still be able to maintain most of the elements inside the CT image. |
|---|