Site qualification standards for Tc-99m-SPECT perfusion imaging in a multi-centre study of MITNEC (Medical Imaging Trials Network of Canada)

Objectives Accuracy of Tc-99m SPECT is being compared to non-nuclear imaging modalities to identify flow-limiting coronary artery disease in patients with symptoms of myocardial ischemia. Phantom qualifying studies established accuracy of SPECT cameras at 6 participating centres for detection of mild ischemia.

Methods Standardized rest-stress ECG-gated Tc-99m scans were acquired on 13 cameras using a torso phantom with left ventricle (LV) cardiac insert. Stress-induced ischemia was simulated with a 3 cc transmural block in the inferior wall (~3%LV mass). Defects were scored in 17 segments using: 1) automated QGS+QPS ischemic Total Perfusion Deficit (iTPD) and 2) visual interpretation of sum difference scores (SDS). In each vascular territory SDS ≥ 2 and iTPD ≥ 2% were considered positive for ischemia. Visual assessment of inferior-wall ischemia in the right coronary artery (RCA) was the qualifying standard. Sensitivity and specificity were determined for automated scores across all vascular territories. Normal variability of stress - rest LVEF, cavity volume (TID) and perfusion defect (Ischemia) values was assessed as mean ± SD.

Results All cameras passed the qualifying scan; unblinded visual scoring confirmed expected true-positive ischemia in the RCA and no false-positive ischemia in other regions. Automated scoring sensitivity and specificity were 69% and 88%, improving to 77% and 88% using an optimized threshold of iTPD ≥ 1.8%. Variability of other diagnostic measures was: LVEF = -0.4 ± 3%, TID = 1.0 ± 0.04, Ischemia = 6 ± 3 %LV across all cameras.

Conclusions Mild stress-induced ischemia was detected accurately on all SPECT systems using visual interpretation, the primary analysis method for this MITNEC study. Sensitivity of automated scoring was improved slightly using an optimized ischemic threshold. Normal variability in diagnostic measures of LVEF, TID and Ischemia should be considered when interpreting patient scans.