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

Abstract

The objective of this study was to determine the impact on detection accuracy of coronary artery disease (CAD) from different combinations of compensation for attenuation, scatter, and resolution using unified normal databases with polar-map quantitation for detection. We studied 102 individuals who either underwent X-ray angiography (57) or were determined to be 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, OH) SPECT camera was used for all acquisitions. The standard filtered backprojection reconstruction (FBP) reconstruction with no attenuation or scatter compensation was performed on the emission data using the 180/spl deg/ data from RAO to LPO. A Butterworth pre-filter of order 5 with a cutoff of 0.25 cycles per pixel was used. Emission data were also reconstructed through 360/spl deg/ using an ordered-subset expectation-maximization (OSEM) algorithm. We used 15 subsets of four angles each with one iteration employed for attenuation compensation (AC) alone, and for AC in combination with scatter compensation (SC). Resolution compensation (RC) was included with the previous compensation methods using 5 iteration of OSEM. Three-dimensional post-reconstruction Gaussian filtering was performed using a standard deviation (sigma) of 0.75 pixels. For the detection of CAD, a progressive improvement in the detection accuracy was observed from FBP without added compensation to OSEM with AC, SC, and RC included. This trend was not consistently observed when comparing detection accuracy for the individual territories, but OSEM with all three compensations had generally the best detection accuracy. Our patient population presented a significant problem with subdiaphragmatic activity due to the large percentage of pharmaceutical stress patients (/spl sim/65%). Therefore, the inferior wall (right coronary artery) territory showed evidence of deterioration in detection accuracy for solely AC as opposed to FBP due to the presence of significant contamination from subdiaphragmatic activity in this patient population.