Value of PET ECG gating in a cross-validation study of cardiac function assessment by PET/MR imaging

Figures

• 

  Figure 1. ECG gating methods for cardiac PET imaging. ECG signal sample measured during the exam (black line) with the detected R peaks by the PET/MR system (blue dots). Cardiac phases definition (gates) based on two approaches: fixed gate duration (FW gating [upper blue grid]), and fixed number of gates (STD and STD-BR gating [lower red grid]) per cardiac cycle

• 

  Figure 2. Upper row: ECG signal (black line) sample with the detected R peaks by the PET/MR system (blue dots) and R-R intervals considered by the acceptance window (red dash line). Abnormal R-R intervals (longer R-R intervals due to arrhythmias and artefacts in R peak detection) are not used in PET reconstruction in STD-BR ECG gating method. Lower row: R-R intervals histogram with the selected window acceptance (dash red line) used in STD-BR

• 

  Figure 3. General scheme of data acquisition, cardiac PET reconstruction and PET/MR comparison of left ventricular volumes and ejection fraction. Cardiac MR, cardiac magnetic resonance; PET, positron emission tomography; AC, attenuation correction; ECG, electrocardiogram; EDV, end-diastolic volume; ESV, end-systolic volume; EF, ejection fraction; LV, left ventricle; FDG, Fluorine-18 fluorodeoxyglucose; STD, standard ECG gating method; STD-BR, standard - beat rejection ECG gating method; FW, fixed width ECG gating method

• 

  Figure 4. Correlation analysis of LV volumes and ejection fraction obtained with MR and PET. PET ESVs correlated better with MRI than PET EDVs. In overall, there were high correlations (Pearson’s r > 0.75), with STD-BR method with the best while STD yielded the poorest correlation. Identity line in red dashed line, and linear regression model in blue solid line

• 

  Figure 5. Bland–Altman plots of LV volumes and ejection fraction obtained with MR and PET modalities. There were slight to moderate underestimation by PET with wide limits of agreement. STD showed the strongest disagreement. No marked differences between STD-BR and FW methods