de-noising spect images from a typical collimator using wavelet transform

introduction: spect is a diagnostic imaging technique the main disadvantage of which is the existence of poisson noise. so far, different methods have been used by scientists to improve spect images. the wavelet transform is a new method for de-noising which is widely used for noise reduction and quality enhancement of images. the purpose of this paper is evaluation of noise reduction in spect images by wavelet. material and methods: to calculate and simulate noise in images, it is common in nuclear medicine to use monte carlo techniques. the simind software was used to simulate spect images in this research. the simulated and real images formed using the current typical (hexagonal) collimator were de-noised by different types of wavelets. results: the best type of wavelet was selected for spect images. the results demonstrated that the best type of wavelet in the simulated and real images increased signal to noise ratio (snr) by 33% and 45% respectively. also, coefficient of variation (cv) decreased by 77% and 71% respectively, while contrast of recovery (cr) was reduced by only 4% and 9% respectively. conclusion: comparing the results for real spect images in this paper with previously acquired results in real pet images, it can be concluded that the images of both nuclear medicine systems using wavelet transform differ in snr and cr by only 5% and 7% respectively, and in cv by about 20%. therefore, wavelet transform is applicable for nuclear medicine image de-noising.