Filtered back projection reconstruction research based on Gaussian in PET images

Positron Emission Tomography (PET) is one of the advanced noninvasive diagnostic techniques of modern medical science. Owing to the reason that PET can display the biological and physiological process by imaging the spatial distribution of radioisotope labeled chemical compounds injected into human body which can help doctor detect disease early, it has become one of the most important parts of modern nuclear medical diagnosis. But the resolution of PET image is rather low, so the improvement of image reconstruction algorithm becomes essential. Filtered back projection (FBP) is a widely used image reconstruction algorithm owing to the characteristics of simplicity and being easy to implement in clinical. However, if the projection data contains noise, the quality of the reconstructed image will degrade badly. To solve this problem in a certain sense, this paper presents a new local Gaussian FBP algorithm. By comparison with R-L and S-L filter, a conclusion can be drawn as follows: the improved FBP algorithm can suppress noise effectively in a certain sense, causing a smaller aliasing in image edge. In addition, the image reconstructed by the new method has a better grayscale contrast and smooth degree.