Aarhus University Seal / Aarhus Universitets segl

Automatic extraction of left ventricular mass and volumes using parametric images from non-ECG-gated 15O-water PET/CT

Research output: Contribution to conferenceConference abstract for conferenceResearchpeer-review

  • J Nordström, Nuclear Medicine & PET, Uppsala, Sweden
  • Hans Harms
  • Mark Lubberink, Nuclear Medicine & PET, Uppsala, Sweden
  • Lars Poulsen Tolbod
  • J van den Berg, Uppsala Univ, Uppsala University, Dept Med Chem, Platform Preclin PET, Sweden
  • T Baron, Department of Cardiology, Uppsala University Hospital, Sweden
  • F.A. Flachskampf, Department of Cardiology, Uppsala University Hospital, Sweden
  • Tanja Kero, Nuclear Medicine & PET, Uppsala, Sweden
  • Jens Sørensen
Introduction: 15O-water positron emission tomography (PET) is considered the gold standard for non-invasive quantification of myocardial blood flow (MBF). It has been shown to identify patients with significant coronary artery disease (CAD) with high accuracy. Hypertrophy with or without dilatation of the myocardium together with a reduction in stroke volume are common in nearly any advanced cardiac disease. Thus, in addition to MBF, assessment of left-ventricular (LV) volumes and mass provides important diagnostic and prognostic information and are vital in the clinical setting. Therefore, the aim of the present study was to investigate the feasibility of measuring LV geometry using dynamic 15O-water PET/CT without ECG-gating. Methods: Parametric images of MBF, perfusable tissue fraction (PTF) and LV blood pool were generated automatically using kinetic modelling. Segmentation of the LV wall using PTF images generated regional MBF and PTF, and total ROI volume which was converted to LV mass (mLV). Segmentation of blood pool images generated LV end-diastolic (EDV) and end-systolic (ESV) volumes. Segmentation thresholds were studied analytically in a subset of patients. The generated values were combined to measure stroke volume (SV=EDV-ESV) and ejection fraction (EF=SV/EDV). Accuracy was determined by comparing PET to cardiac magnetic resonance (CMR) in 30 asymptomatic patients with high grade LV regurgitation (group A). Precision was determined as inter-observer variation in group A and by a test-retest PET study in 15 patients with suspected ischemic heart disease at rest (group B). Normal values were defined in nine healthy volunteers. Results: The accuracy of PET was good with correlations of r= 0.83, 0.88, 0.86 and 0.76 for mLV, EDV, ESV and SV, respectively, compared to CMR (P<0.001 for all). EF was in the normal range (53%-75%) and had a weaker correlation of r=0.51 (P=0.004). mLV was overestimated by 18%±13% compared to CMR and the residuals were closely related to the eccentric remodelling (mLV/EDV) defined by CMR. EDV, ESV and SV were underestimated by 10%±10%, 19%±13%, 3%±15%, respectively, compared to CMR. EF was overestimated by 4%±5% compared to CMR. Inter-observer correlation and test-retest repeatability were excellent (r>0.97). Normal values from PET were in line with normal values from a local CMR cohort. Conclusion: Microvascular function and LV geometry can be measured automatically with high accuracy and precision from a single non-gated 15O-water PET. The accuracy of mLV from PET is affected by LV eccentricity.
Original languageEnglish
Publication year2017
Publication statusPublished - 2017
EventEuropean Association of Nuclear Medicine Annual Congress - Vienna, Austria
Duration: 21 Oct 201725 Oct 2017


ConferenceEuropean Association of Nuclear Medicine Annual Congress
Internet address

See relations at Aarhus University Citationformats

ID: 120302923