Innovative procedure for measuring left ventricular ejection fraction from 18F-FDG first-pass ultra-sensitive digital PET/CT images: evaluation with an anthropomorphic heart phantom

Figures

• 

  Fig. 1. Schematic depiction of the protocol used to determine the LVEF in the heart-phantom pair. CT, computed tomography; LV, left ventricle; LVEF, left ventricular ejection fraction; MRI, magnetic resonance imaging; QBS, quantitative blood pool SPECT; PET, positron emission tomography; pERNA, planar equilibrium radionuclide angiocardiography; SPECT, single-photon emission computed tomography

• 

  Fig. 2. Schematic depiction of the automated Python method for producing “pseudo-planar” PET images. The “pseudo-planar” image was obtained by summing the signals in all 127 PET slices along the z-axis. Preorientation allowed the z-axis to be aligned with the cardiac small axis. LV, left ventricle; RV, right ventricle

• 

  Fig. 3. The 3D-printed heart-phantom pair showing the heart at end-diastole and end-systole. The pipes on the top of the phantoms were used to fill the left and right cavities and the myocardium compartment with water or radionuclide solutions

• 

  Fig. 4. Images of the end-diastole/end-systole heart-phantom pair. For the 3D modalities, namely, magnetic resonance imaging (MRI), single-photon emission computed tomography (SPECT), and native positron emission tomography (PET), only the central slice in the cardiac short axis is shown. The SPECT, planar scintigraphy, and “pseudo-planar” PET images show representative examples of manual segmentation. The “pseudo-planar” PET images along the cardiac small axis had better image quality (especially in terms of spatial resolution) than the equivalent planar scintigraphy images. LV, left ventricle; RV, right ventricle

• 

  Fig. 5. Left ventricle A volumes and B ejection fractions of the heart-phantom pair. These values were measured by MRI, SPECT, planar scintigraphy, native 3D PET/CT, and “pseudo-planar” PET/CT projections on the cardiac small axis and horizontal long axis. LVEF, left ventricular ejection fraction. The prefixes “a” and “m” signify automated and manual calculations, respectively. Left ventricle end-diastole and end-systole volumes are not shown in A for planar scintigraphy or the “pseudo-planar” PET/CT projections because these modalities do not provide information about left ventricle volumes. The automated left ventricle volumes and LVEF values shown for SPECT and native PET were calculated by using the surface-based QBS method. 1“Full automated” computations represent analysis with no human intervention, so no variability expected nor observed. 2“Semi-automated” processes stand for automated computation of LVEF after a manual placement of anatomical landmarks (valves and septum). For semi-automated and manual processes, the value corresponds to the average of 4 repetitions of the image treatment with 4 different operators, and the error bar represents the standard deviation…

• 

  Fig. 6. Automated segmentation with quantitative blood pool SPECT (QBS) of native 3D PET/CT images at end-diastole. The lines indicate the imprecise ventricle edges that were determined by this automated segmentation algorithm

• 

  Fig. 7. Automated segmentation of “pseudo-planar” PET/CT images summed along the cardiac short axis