comparision of DFE - based , gaussian and wavelet approaches for enhancing PET images

Abstract

Abstract: The ability to diagnose many of the severe diseases, such as cancer and heart problems, is a challenge for physicians and radiologists. Recently, a wide range of technologies are being used to increase the accuracy of physicians’ diagnoses. Positron Emission Tomography (PET) is an important nuclear medicine imaging technique which enhances the effectiveness of diagnosing many diseases. The raw-projection data, i.e. the sinogram, from which the PET is reconstructed, contains a very high level of Poisson noise. Radiologists face difficulties when reading and interpreting PET images because of the high noise level. The later may lead to erroneous diagnoses. Finding a suitable de-noising technique for PET images has attracted many researchers in the last two decades as this can significantly alleviate the problem. In this work, we compare and investigate four type of filters for enhancing PET image: The Perona and Malik, construction curvature motion, Gaussian, and wavelet fillers, The group of our research used matlab for implementing and running the filters then we will make a quantitative comparison using three measuring approach PSNR, NR, and correlation.