Determination of optimum filter in myocardial SPECT: A phantom study

In myocardial perfusion SPECT imaging, images are degraded by photon attenuation, the distance-dependent collimator, detector response and photons scatter. Filters greatly affect quality of nuclear medicine images A phantom simulating heart left ventricle was built. About 1mCi of 99mTc was injected into the phantom. Images were taken from this phantom. Some filters including Parzen, Hamming, Hanning, Butterworth and Gaussian were exerted on the phantom images. By defining some criteria such as contrast, signal to noise ratio, and defect size detectability, the best filter can be determined. 0.325 Nyquist frequency and 0.5 Nq was obtained as the optimum cut off frequencies respectively for hamming and hanning filters. Order 11, cut off 0.45 Nq and order 20 cut off 0.5 Nq obtained optimum respectively for Butterworth and Gaussian filters. The optimum member of every filter’s family was obtained.