[Experimental and clinical studies of frequency domain Butterworth filter and wiener filter on process of radionuclide imaging]

A frequency domain filtering method provides accurate frequency response. In order to provide optimal radionuclide image quality, Butterworth or Wiener filters were applied in the frequency domain. A method is developed to determine automatically their optimal parameters from the power spectra of the object images. Phantom studies showed that proper parameters were derived by this method. Thirty-one cases of multigated cardiac blood-pool images and phantom images modeled on the cardiac images were filtered by 3 x 3 points temporal smoothing, frequency domain Butterworth and Wiener filters. From these images, ejection fractions were calculated and evaluated the correlation between the actual ejection fractions in the phantom study or the values derived from the contrast left ventriculography. The better correlation coefficients were derived with the frequency domain Butterworth filter than with the temporal smoothing. And more desirable correlation coefficients were derived with the frequency domain Wiener filter than with the frequency domain Butterworth filter, especially in the low count studies. Butterworth or Wiener filtering in the frequency domain reduces noise and improves the capacity for quantitative analysis in the radionuclide images.