Aarhus University Seal / Aarhus Universitets segl

Kirsten Bouchelouche

Automatic extraction of myocardial mass and volumes using parametric images from dynamic nongated PET

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisTidsskriftartikelForskningpeer review

DOI

Dynamic cardiac positron emission tomography (PET) is used to quantify molecular processes in vivo. However, measurements of left-ventricular (LV) mass and volumes require electrocardiogram (ECG)-gated PET data. The aim of this study was to explore the feasibility of measuring LV geometry using non-gated dynamic cardiac PET. METHODS: Thirty-five patients with aortic-valve stenosis and 10 healthy controls (HC) underwent a 27-min 11C-acetate PET/CT scan and cardiac magnetic resonance imaging (CMR). HC were scanned twice to assess repeatability. Parametric images of uptake rate K1 and the blood pool were generated from non-gated dynamic data. Using software-based structure recognition the LV wall was automatically segmented from K1 images to derive mLV and wall thickness (WT). End-systolic (ESV) and end-diastolic (EDV) volumes were calculated using blood pool images and used to obtain stroke volume (SV) and LV ejection fraction (LVEF). PET measurements were compared with CMR. RESULTS: High and linear correlations were found for LV mass (r=0.95), ESV (r=0.93) and EDV (r=0.90), and slightly lower for SV (r=0.74), LVEF (r=0.81) and WT(r=0.78). Bland Altman analyses showed significant differences for mLV and WT only and an overestimation for LVEF at lower values. Intra- and inter-observer correlations were >0.95 for all PET measurements. PET repeatability accuracy in HC was comparable to CMR. CONCLUSION: LV mass and volumes are accurately and automatically generated from dynamic 11C-acetate PET without ECG-gating. This method can be incorporated in a standard routine without any additional workload and can, in theory, be extended to other PET tracers.
OriginalsprogEngelsk
TidsskriftJournal of Nuclear Medicine
Vol/bind57
Nummer9
Sider (fra-til)1382-87
Antal sider6
ISSN0161-5505
DOI
StatusUdgivet - 28 apr. 2016

Se relationer på Aarhus Universitet Citationsformater

ID: 99706361