Assessment of myocardial perfusion can be made more precise by incorporating semiquantitative scoring.
Quantitative analysis of relative myocardial perfusion provides objective and accurate information, with lower variability compared to subjective visual interpretation.
Quantitative analysis can be improved by incorporating multiple positions, attenuation correction imaging, and motion correction.
Nonperfusion imaging parameters can also be assessed quantitatively to improve diagnostic and prognostic accuracy.
Artificial intelligence is emerging as a tool to improve image processing and reconstruction and to optimize disease diagnosis and risk estimation.
Single-photon emission computed tomography (SPECT) and positron emission tomography (PET) myocardial perfusion imaging (MPI) are frequently used for diagnosis or risk prediction in patients with known or suspected coronary artery disease (CAD). There is an abundance of evidence supporting their roles for disease diagnosis and risk stratification in all key patient populations.1–5 While visual inspection of regional perfusion can frequently identify important abnormalities, methods to quantify the extent and severity of those abnormalities can be used to improve the diagnostic and prognostic value.6,7 We will review approaches to quantify regional myocardial perfusion abnormalities, both with visual semiquantitative scores and in an automated fashion. Additionally, we will provide an overview of techniques to improve the quantification by considering multiple images and applying motion correction. Next, we will review fully automated quantification of selected nonperfusion variables including left ventricular ejection fraction (LVEF) and transient ischemic dilation (TID). Quantification of absolute myocardial blood flow (MBF) for PET, including potential clinical roles, and emerging evidence for similar measurements with SPECT will also be discussed. While the standard methods for quantitation have already significant advanced SPECT and PET MPI interpretation, artificial intelligence (AI) is emerging as an important next step for the field. In the final section, we review current evidence for the use of AI for PET and SPECT MPI, including potential clinical applications.
VISUAL SEMIQUANTITATIVE SCORING
Initial attempts at quantifying regional perfusion abnormalities on SPECT and PET relied on expert visual interpretation.6 Ladenheim et al. were the first to demonstrate, in a study of 1689 patients, the additive prognostic value of extent and severity of myocardial hypoperfusion.6 This finding formed the basis for semiquantitative scoring of myocardial perfusion, as shown in Figure 15-1.
Semiquantitative scoring of myocardial perfusion. Graphical representation of a perfusion polar map, with 17-segments numbered according to the current AHA model. Each segment is scored from 0 (normal) to 4 (absent uptake of radioactivity). The total score for all segments on stress perfusion is the summed stress score. The percent myocardium involved at stress is calculated from the total score divided by the maximum potential score (17 × 4 = 68) and multiplying by 100.7 The same process can be repeated at rest to calculate summed rest score. Summed difference score is the difference between summed stress and rest scores. Segments are assigned to the appropriate ...