Assessment of cardiac single-photon emission computed tomography performance using a scanning linear observer

Area of Science:

• Medical Imaging
• Nuclear Cardiology
• Systems Engineering

Background:

• Single-photon emission computed tomography (SPECT) is crucial for diagnosing myocardial ischemia and infarction.
• Optimizing SPECT system design is essential for enhancing the detectability of cardiac defects.

Purpose of the Study:

• To assess and compare the performance of three GE cardiac SPECT systems (GE27, GE19, GE13) with varying detector counts.
• To evaluate the utility of the scanning linear observer (SLO) and LROC/EROC curves for assessing SPECT system design.

Main Methods:

• Employed the scanning linear observer (SLO) model, which analyzes projection data for hardware performance assessment independent of reconstruction algorithms.
• Utilized localization receiver operating characteristic (LROC) and estimation receiver operating characteristic (EROC) curves to evaluate defect location and extent estimation.
• Assessed three GE Discovery NM 530c based systems: GE27 (27 detectors), GE19 (19 detectors), and GE13 (13 detectors).

Main Results:

• The GE19 system demonstrated performance metrics (AUEC: 0.8488, TPF at 5% FPF: 76.46%) very close to the GE27 system (AUEC: 0.8545, TPF at 5% FPF: 77.1%).
• The GE13 system showed slightly lower performance (AUEC: 0.8329, TPF at 5% FPF: 73.55%).
• The GE19 system achieved comparable results to the GE27 system while utilizing eight fewer detectors, representing a 30% reduction.

Conclusions:

• The combination of SLO and LROC/EROC curves provides a robust method for evaluating cardiac SPECT systems.
• The GE19 system configuration offers a compelling balance of performance and cost-effectiveness by reducing detector count without significant loss of diagnostic information.
• This approach aids in determining optimal SPECT system configurations for improved cardiac defect detection.