The Yale-CQ method for cardiac SPECT quantification: phantom validation and patient application

Performance of the Yale-CQ SPECT quantification method was evaluated using phantom and patient data. Forty-two phantom experiments and 218 patient studies were included in this evaluation. The quantification was carried out by the automatic and manual methods. For the phantom data, the success rate (SR) of the automatic method was 100%. The mean absolute errors of estimating the center location (MAE-x, MAE-y and radii MAE-r) of the short axial slices were 0.28, 0.57 and O.11 pixels, respectively, as for comparison of the automatic and manual methods. Correlation between the actual defect size and the average defect score was excellent (r=0.96). The consistency of the results from the automatic and manual methods was high (r=0.98). Inter-observer results obtained by the manual method were highly reproducible (r=0.99). For patient data, SR was 76%, and MAE-x, MAE-y and MAE-r were 0.72, 1.08 and 1.03 pixels, respectively. Results by the automatic and manual methods were highly consistent (r=0.99). This validation showed the Yale-CQ method may provide a reliable measurement of myocardial perfusion abnormality which can be expressed as a percent of LV