Pitfalls in quantitative myocardial PET perfusion I: Myocardial partial volume correction

Abstract

Background. PET quantitative myocardial perfusion requires correction for partial volume loss due to one-dimensional LV wall thickness smaller than scanner resolution. Methods. We aimed to assess accuracy of risk stratification for death, MI, or revascularization after PET using partial volume corrections derived from two-dimensional ACR and three-dimensional NEMA phantoms for 3987 diagnostic rest-stress perfusion PETs and 187 MACE events. NEMA, ACR, and Tree phantoms were imaged with Rb-82 or F-18 for sizedependent partial volume loss. Perfusion and Coronary Flow Capacity were recalculated using different ACR-and NEMA-derived partial volume corrections compared by Kolmogorov-Smirnov statistics to standard perfusion metrics with established correlations with MACE.

Reliability and Reproducibility of Absolute Myocardial Blood Flow: Does It Depend on the PET/CT Technology, the Vasodilator, and/or the Software?

“…While widely reported [ 9 •, 10 ], global CFR or stress perfusion alone, or perfusion within fixed regions of interest (ROIs) for assumed standard or angiogram defined arterial distributions, average the perfusion of different neighboring arterial distributions within each arbitrary ROI [ 1 , 2 , 3 •, 4 •, 5 •, 6 ••, 7 ••, 11 •, 12 ]. Such arbitrary ROIs preclude tracking severity of iso-contour boundaries for precise size and severity of artery-specific stenosis done in Fig.…”

“…Maximum activity along 64 radii in each of 21 short-axis tomographic slices provides μCi/g of myocardium with maximal statistical certainty, as shown in Fig. 1e [ 1 , 2 , 3 •, 4 •, 5 •, 6 ••, 7 ••, 11 •, 12 – 14 ]. Pixel data allow size and severity of arterial distribution as it actually is for either severe abnormalities as seen in Fig.…”

“…Test-retest precision (coefficient of variance—COV) in the same patient was determined by serial quantitative imaging minutes and days to weeks apart in 100 healthy young volunteers with no risk factors and in 120 volunteer patients with risk factors or known CAD [ 11 •, 12 – 15 ]. The test-retest COV for serial stress ml/min/g minutes apart is ± 10%.…”

“…The HeartSee flow model (510K-171303) with a 2-min arterial phase image and 5-min myocardial image for Rb-82 and N-13 ammonia is validated experimentally [ 20 , 21 ] and clinically as outlined above [ 1 , 2 , 3 •, 4 •, 5 •, 6 ••, 7 ••, 11 •, 12 – 15 , 17 – 21 ] for full-dose Rb-82 or N-13 ammonia with the appropriate flow model for each and longer myocardial phase images for N-13 ammonia. Direct objective comparison to multi-compartmental models by another PET center concluded that the retention model of HeartSee flow software provided for Rb-82—“higher sensitivity for detection and localization of abnormal flow and myocardial perfusion reserve…without the computational complexity and sensitivity to noise – of the multi-compartmental model” [ 23 ].…”

“…However, 3D acquisition on BGO 2D-3D scanners fails to acquire accurate arterial input over minutes-long single arterial phase images due to inadequate corrections for random coincidences, dead time, and scatter [ 11 •, 12 ]. As a feasibility study for these BGO 2D-3D scanners, serial short 15-s images each corrected separately then summed for a 2-min arterial phase image provide adequate quality arterial phase images on which optimal ROI for arterial input can be determined [ 11 •, 12 ].…”