Comparison of detection of perfusion defects in SPECT for different reconstruction strategies using polar map quantitation

The objective of this work was to extend our study into the effects of attenuation and scatter compensation on the apparent distribution of Tc-99m sestamibi in cardiac slices to include compensation for resolution. Furthermore, we also determine the accuracy of diagnosing CAD using polar map quantitation with a unified normal database. We studied 102 individuals who either underwent X-ray angiography (57) or were determined to be in a low risk group (less than 5% likely) to have CAD (45). The low likelihood group was identified using standard criteria. Both groups underwent stress testing (physical or pharmacological) before imaging commenced. A Philips Medical Systems Prism 3000 (Philips Medical Systems, Cleveland Ohio) SPECT camera was used for all acquisitions. The standard FBP reconstruction with no attenuation or scatter compensation was performed on the emission data using the 180/spl deg/ data from RAO to LAO, and a Butterworth pre-filter of order 5 and cutoff of 0.25 1/pixel. Emission data were also reconstructed through 360/spl deg/ using an ordered-subset expectation-maximization (OSEM) algorithm. In our implementation, we used 15 subsets of four angles each and one iteration employing attenuation compensation (AC) alone, and in combination with scatter compensation (SC). Resolution compensation was included with the previous compensation methods (RC) using 5 iteration of OSEM. Three-dimensional post-reconstruction Gaussian filtering was done using a sigma of 0.75 pixels. As expected, FBP do poorly with a unified normal database when inferior and anterior regions are affected. The three methods that included compensation for attenuation were highly susceptible to extracardiac activity, influencing the maximum count in the polar map and change diagnoses of normal into CAD. This was evident in the areas under the ROC curves for all methods.