Interdependence between measures of extent and severity of myocardial perfusion defects provided by automatic quantification programs

Abstract

To evaluate the accuracy of the values of lesion extent and severity provided by the two automatic quantification programs AutoQUANT and 4D-MSPECT using myocardial perfusion images generated by Monte Carlo simulation of a digital phantom. The combination between a realistic computer phantom and an accurate scintillation camera simulation tool allows the generation of realistic single-photon emission computed tomography (SPECT) images similar to those obtained in clinical patient studies. The NCAT phantom and the SIMIND Monte Carlo program were used to simulate myocardial perfusion studies. Perfusion defects with sizes ranging from 5 to 17% of the left ventricular wall volume and reductions in tracer uptake of 20, 60 and 100% were simulated in three vascular territories. The values of the extent provided by the programs were dependent on the reduction in tracer uptake, i.e. the severity. Similarly, the measures of severity were dependent on the size of the lesions. The severity provided by AutoQUANT for different defects was not dependent on the location, whereas 4D-MSPECT presented different values depending on the location in the left ventricle. The measures of extent and severity of the defects with the same true extent and activity uptake reduction provided by the two programs were different. The NCAT phantom and the SIMIND Monte Carlo program were shown to be useful in simulating clinical myocardial SPECT studies. The quantification programs gave values of lesion extent that were dependent on the magnitude of the severity. Users should therefore consider this dependence when interpreting results from these programs.