Evaluation of SPECT system performance by quantatively comparing planar and reconstructed spatial resolutions

Theoretically, the spatial resolution of the SPECT reconstructed image is determined by the intrinsic resolution of the detector, the collimator, the sampling bin width and the filter function used. However, in a practical SPECT system, the spatial resolution may be further degraded by the center of rotation error, spatial distortion, collimator defects, reconstruction software and computer hardware. The center of rotation error results in misregistration of the projection data, and for small errors, introduces spatial resolution degradation of approximately two times the misregistration error. Two SPECT systems were evaluated for the effect of misregistration on spatial resolution. Thirty-six projection images of a Tc-99m line source taken between 0 and 360 degrees were fitted to a sinusoid to determine the maximum error epsilon/sup max between the measured and fitted data. Planar and reconstructed SPECT images of line sources placed on the axis of rotation and with a radius of rotation equal to 15 cm were acquired using low energy all purpose collimation. The SPECT images were reconstructed using a generalized Hann filter with a cut-off frequency greater than four times the Nyquist frequency. The FWHM(mm) of the planar and reconstructed images of the line sources were determined. Results from the firstmore » (second) SPECT systems were: /epsilon/sup max = 0.25 mm(0.8mm), FWHM/sub planar/ = 14.7mm(15.0mm), and FWHM/sub SPECT/ = 14.9 mm(16.3mm). It is concluded that the quantitative comparison of planar and reconstructed spatial resolutions provide an accurate evaluation of center of rotation errors.« less