Two optimal prefilter cutoff frequencies needed for SPECT images of myocardial perfusion in a one-day protocol

The purpose of this study was to determine if cutoff frequencies of two-dimensional Butterworth filters need to be changed, depending upon the amount of radiopharmaceutical administered in one-day protocols for reconstructing SPECT images of myocardial perfusion.

A scintilation camera and a phantom were used with 2.2 MBq and 10.6 MBq 99mTc. These activities corresponded to the approximate amounts used in our one-day myocardial perfusion imaging protocol. The projection data were collected in 30 positions spread over 180 degrees. Thirty- and 150-sec acquisitions per position were performed to obtain the study and the high-quality reference images. Thirteen cutoff frequencies of the filter were used to reconstruct the SPECT images. Optimal cutoff frequencies were determined by visual assessment and by calculating the normalized mean square error (n.m.s.e). These results were then compared. The same protocol was used to image three volunteers with no heart disease and the images were evaluated visually.

In the phantom study, the optimal cutoff frequencies were 0.225 cycles/pixel for 2.2 MBq, and 0.275 cycles/pixel and 0.25 cycles/pixel for 10.6 MBq by visual inspection, and were 0.225 cycles/pixel and 0.275 cycles/pixel by calculating the n.m.s.e., respectively. In the patient study, the lower cutoff frequency of 0.20 cycles/pixel was optimal for the exercise study (low dose), while the higher cutoff frequencies of 0.275 cycles/pixel was optimal for the resting study (high dose).

The optimal cutoff frequencies differed for the two activity levels. Cutoff frequencies should be changed according to the amounts of radiopharmaceuticals used for the first and the second examinations in one-day protocols.