Absolute quantification for small-animal PET

Abstract

Abstract-Quantification is important in preclinical PET studies. To achieve absolute quantification, an accurate reconstruction algorithm is necessary. Such an algorithm includes corrections for different effects such as geometric sensitivity of the scanner, detection efficiency, attenuation, scatter and random coincidences. In this work we present a method for performing absolute quantification on the LabPET system. All acquisitions were done on a GE Triumph system. This tri-modality system consists of a micro-PET (LabPET), micro-CT (X-O) and micro-SPECT (X-SPECT) scanner. Three PET scans were done. In the first scan 5 vials with different activity concentrations of 18 F-FDG were scanned. The total activity inside the scanner was 80 MBq. The second scan was performed after 4 hours when the total activity in the scanner had decayed to 20 MBq. In the third scan 3 vials and 1 sphere were scanned with a total activity of 20 MBq. Before each PET scan a micro-CT scan was acquired.Point sources with a known activity were placed inside the field of view. The counts obtained in these point sources are used to obtain a correction factor for absolute sensitivity. Reconstruction was done using a 3D ML-EM reconstruction with micro-CT based attenuation correction. VOIs were drawn over the vials and the sphere in the reconstructed images. The total activity in the VOIs was calculated using the correction factor for absolute sensitivity. It was compared to the activity measured in a dose calibrator.The average quantification error was 56 %, 6.4 % and 0.6 % for the first, second and third scan. The high error in the first scan is explained by count rate effects, as 80 MBq can be considered a high activity level for this system.The feasibility of absolute quantification on the LabPET system was demonstrated. When the count rate is below 20 MBq absolute quantification is possible with an average quantification error smaller than 6.4 %.