Automated quantification of 99mTc sestamibi myocardial perfusion compared with visual analysis.

Objectives: The visual interpretation of 99mTc sestamibi single photon emission computed tomography (SPECT) myocardial perfusion images can be challenging due to the quantity of scan information generated, the large number of normal variants, attenuation artifacts and gender differences. The development of automated, computer derived, quantitative indices of perfusion can assist in this interpretation by providing an objective measure. It is important to verify that similar results can be obtained when the software is used in centres outside those where the algorithms were initially developed. Our objective was to assess the degree of concordance between the visual and automated diagnostic assessments of 99mTc sestamibi SPECT.

Methods: We studied 718 patients referred for 99mTc sestamibi SPECT myocardial perfusion imaging. The SPECT studies were initially interpreted visually without benefit of computer based analysis, and were then subjected to blinded reprocessing to extract quantitative indices of perfusion.

Results: There was very good agreement between the visual and quantitative diagnostic classifications. When a visual abnormality was taken to be the reference standard, the automated summed stress score (SSS) showed agreement (SSS>3) in 80% (kappa 0.60, P<0.0001). The area under the receiver operating characteristic (ROC) curve was 0.89 (95% confidence interval (CI), 0.86-0.91). Concordance was greater in those with previous myocardial infarction or severe perfusion defects, but was not affected by age, prior revascularization, stress procedure or heart rate. Concordance over the presence or absence of visual reversibility and the summed difference score (SDS) in abnormal scans was slightly lower (overall agreement 73% (kappa 0.36, P<0.00001) and ROC area 0.84 (95% CI, 0.77-0.90)).

Conclusion: Automated quantification of 99mTc sestamibi SPECT myocardial perfusion with the SSS and SDS provides objective diagnostic information and concordance when compared with conventional visual image interpretation.