Abstract
372
Objectives Routine clinical use of dynamic PET/CT imaging is hampered by long scan times and difficulties with obtaining accurate arterial input functions. Long scan times are usually required to collect the appropriate imaging data for use in models while invasive blood sampling is needed to properly scale and correct imaging data for various effects. This work examines the use of an external radiation detector system (Lucerno Dynamics, LLC) to non-invasively estimate an arterial input function for use in whole-body continuous bed motion (CBM) dynamic imaging and subsequent late time-point Patlak analysis.
Methods Small external detector modules were placed on the left and right bicep of each patient (near the cephalic or basalic veins) as well as the right carotid artery while seated in the injection chair. Data collection began approximately 20 seconds prior to IV injection of FDG. Data were collected for approximately 4 minutes before moving patients to the holding area to complete their remaining uptake time and data acquisition. The IV was not removed so veinous blood could be drawn and counted 40 minutes post-injection to estimate arterial blood activity concentrations (1). After a 45 minute uptake period, all detectors were removed and patients moved to the PET/CT for imaging. A dynamic multi-bed CBM acquisition was performed with 6 passes collected over a total acquisition time of approximately 15 minutes. Data were reconstructed and time activity curves generated from the external detector system (fig. 1) and PET/CT imaging. Modeling and calculation of metabolic rate of glucose (MRGLu) was performed using Patlak analysis of brain regions of interest with the input function derived from triple exponential fit of the data collected by the external detector device. Detector and image data were scaled by veinous blood sample activity concentration measurements. Values were compared to previously published results for healthy brain metabolism.
Results Calculations of MRGlu with this technique resulted in average values of 3.5 mg/ml/100g, corresponding with previously published values for healthy brains (3.34 mg/ml/100g) and differing by only 4% (2, 3). Our measurements from this method also differ by only 7% from previous work by our group using CBM dynamic imaging that yielded values of 3.7 mg/ml/100g (4).
Conclusions The technique described enables routine use of whole-body dynamic imaging requiring only 15 minutes of PET scanner time. This technique also uses a novel method for estimation of arterial input functions for use in quantitative modeling. Further development of this technique may result in routine use of dynamic PET imaging to obtain improved quantitative imaging and more robust data comparison between patients.