Design and Assessment of Cardiac SPECT Systems

Abstract

Single-photon emission computed tomography (SPECT) is a modality widely used to detect myocardial ischemia and myocardial infarction. Objectively assessing and comparing different SPECT systems is important so that the best detectability of cardiac defects can be achieved. Whitaker, Clarkson, and Barrett's study on the scanning linear observer (SLO) shows that the SLO can be used to estimate the location and size of signals. One major advantage of the SLO is that it can be used with projection data rather than reconstruction data. Thus, this observer model assesses overall hardware performance independent by any reconstruction algorithm. In addition, we will show that the run time of image-quality studies is significantly reduced. Several systems derived from the GE CZT-based dedicated cardiac SPECT camera Discovery 530c design, which is officially named the Alcyone Technology: Discovery NM 530c, were assessed using the performance of the SLO for the task of detecting cardiac defects and estimating the properties of the defects. Clinically, hearts can be virtually segmented into three coronary artery territories: left anterior descending artery (LAD), left circumflex artery (LCX), and right coronary artery (RCA). One of the most important functions of a cardiac SPECT system is to produce images from which a radiologist can correctly predict in which territory the defect exists. A good estimation of the defect extent from the images is also very helpful for determining the seriousness of the myocardial ischemia. In this dissertation, both locations and extent of defects were estimated by the SLO, and system performance was assessed using localization receiver operating characteristic (LROC) / estimation receiver operating characteristic (EROC) curves. Area under LROC curve (AULC) / area under EROC curve (AUEC) and true positive fraction (TPF) at specific false positive fraction (FPF) can be treated as the figures of merit (FOMs). As the results will show, a combination of the SLO and LROC / EROC curves can determine the configuration that has the most estimation/detection information and thus is a useful method for assessing cardiac SPECT systems.