Non-Local Means Based Image Enhancement on Coronary Angiography Images

Developing technological infrastructure has enabled the development of computer based biomedical systems as in many areas in the field of medicine. One of these systems are biomedical imaging systems. Many studies are being conducted for a new image processing techniques to improve the performance of this systems. Noise reduction is one of the important steps in biomedical image processing. In this study, noise reduction was performed to emphasize coronary arteries on x-ray heart angiography images. For this purpose, the original angiography images were smoothed using non-local averages. Thus, insignificant groups of pixels in the image described as noise has been removed. Then, the boundaries of coronary arteries are sharpened with first and second derivatives of image based combined enhancement method. It is seen that the mean square error values obtained by the proposed method are more successful when compared with the noise reduction results obtained using the Wiener filter. These results show that the non-local means method can be used as a successful pre-processing method for noise reduction in angiography images.