Quantification of SPECT myocardial perfusion images: Methodology and validation of the Yale-CQ method

Abstract

Background. Quantification of single photon emission computed tomography (SPECT) images is important for reproducible and accurate image interpretation. In addition, SPECT quantification provides important prognostic information. The purpose of this study was to validate the Yale circumferential quantification (Yale-CQ) method in phantom studies. Methods. Myocardial perfusion defects of varying extent and severities were simulated in a cardiac phantom with fillable defect inserts. Forty-five different phantom configurations simulated 45 different myocardial perfusion defect sizes, ranging from 1.6% to 32% of the cardiac phantom volume. Automatic processing was compared with manual processing in the phantom SPECT studies. Results. The automatic Yale-CQ algorithm performed well in all phantom studies. Compared with manual processing, the mean absolute error for automatically determined center of short axis slices was 0.27 pixel in the x direction, 0.45 pixel in the y direction, and 0.15 pixel in radius. Quantification of phantom defects with the Yale-CQ method correlated well with actual defect sizes (R = 0.99), but there was a systematic underestimation (mean error = −7.9%). With derived correction factors the overall correlation between 45 phantom defects and actual defect sizes was excellent, and the estimation error was significantly improved (R = 0.98, mean error = −0.82% for manual method and −0.95% for automatic method). Conclusion. The automatic processing algorithm performs well for the phantom studies. Myocardial perfusion abnormalities can be quantified accurately by use of the Yale-CQ method. Quantified SPECT defect size can be expressed as a percentage of the left ventricle.