Head-to-head comparison of uncorrected and scatter corrected, summed and end diastolic myocardial perfusion SPECT in coronary artery disease.

Objectives: Compared with other non-invasive methods for diagnosing coronary artery disease (CAD), myocardial perfusion imaging (MPI) suffers from some lack of specificity, especially in patients with a small heart. Allowing the assessment of perfusion on end diastolic images instead of summed images, gated single photon emission computed tomography (SPECT) constitutes an attractive method for increasing the accuracy of MPI. Scatter correction, known to improve image contrast, might also be interesting. The present study aimed at comparing scatter corrected to uncorrected gated MPI for CAD diagnosis.

Methods: The results for 100 patients referred for gated 99mTc sestamibi SPECT were analysed. They were divided into two subgroups according to their end systolic volume (ESV) measured by QGS analysis (group A, ESV > or =30 ml, n=65; group B, ESV <30 ml, n=35). For each patient, a total defect score (TDS) was quantified on four polar maps (uncorrected and scatter corrected, summed, and uncorrected and scatter corrected, end diastolic). The optimal TDS separating non-CAD from CAD patients was calculated by analysis of the receiver operating characteristic (ROC) curve for the four data sets, using the coronary angiogram as a 'gold standard'.

Results: In the whole patient population, the accuracy of the uncorrected data was 67% for the end diastolic images and 71% for the summed images (sensitivity, 66% and 70%; specificity, 71% and 79%, respectively). After scatter correction, the accuracy did not change for the end diastolic data (accuracy, 67%; sensitivity, 63%; specificity, 93%) and increased to 74% for the summed data (sensitivity, 73%; specificity, 79%). In group A, the uncorrected data were 72% accurate for the end diastolic images and 78% for the summed images (sensitivity, 72% and 79%; specificity, 75% and 75%, respectively). After correction, the accuracy of end diastolic images increased to 77% (sensitivity, 77%; specificity, 75%), and did not change for the summed images (accuracy, 78%; sensitivity, 79%; specificity, 75%). In group B, the accuracy of uncorrected images amounted to 51% for both end diastolic and summed data (sensitivity, 48% and 40%; specificity, 60% and 80%, respectively). After correction, it increased to 57% for the end diastolic images and to 63% for the summed images (sensitivity, 48% and 64%; specificity, 80% and 60%, respectively).

Conclusions: Despite lower blurring on end diastolic compared with summed images, non-scatter corrected end diastolic data were least accurate for the diagnosis of coronary artery disease in patients with a high prevalence of disease. Scatter correction, by improving the delineation of perfusion defects, increased the accuracy of quantitative MPI for the diagnosis of CAD in a large number of patients, more particularly in those with a small heart.