An automated analysis program for the evaluation of cardiac PET studies: initial results in the detection and localization of coronary artery disease using nitrogen-13-ammonia

Abstract

Positron emission tomography (PET), in combination with myocardial blood flow tracers, allows highly accurate diagnosis of coronary artery disease using visual data interpretation. To increase the objectivity of data analysis and to reduce interobserver variability, we developed an automated analysis method for the three-dimensional definition of myocardial activity, which includes true volumetric data extraction and mathematical constraints of activity sampling to the expected shape of the left ventricle. Data are displayed in a standardized polar map or three-dimensional format for comparison with a normal database. The first clinical evaluation of this method in 52 patients using receiver operating characteristics (ROC) curve analysis demonstrated high diagnostic accuracy for detection as well as localization of coronary artery stenosis in predefined vascular territories. The interobserver and intraobserver agreement for localization of disease was excellent, with correlation coefficients varying from 0.85 to 0.99 for individual vascular territories. Thus, this automated quantitative analysis program provides highly accurate and reproducible evaluation of cardiac PET flow studies. Definite determination of its diagnostic accuracy requires a prospective multicenter trial in a larger patient population employing the criteria for abnormality established in this initial clinical evaluation.