Euromedim 2006 :1st European Conference on Molecular Imaging Technology

Implementation of a methodology for the analysis of small animal PET images

11 May 2006, 14:00

1h

Palais du Pharo, Marseille

poster Molecular Imaging needs for biologists and physicians Poster session : Imaging systems, Molecular Imaging

Speaker

Dr Antonello Spinelli (Policlinico S. Orsola-Malpighi, Bologna, Italy)

Description

Introduction Small animal positron emission tomographs (PET) are becoming extremely popular allowing preclinical investigation of new drugs, new radiotracer etc. by performing several sequential scans on the same animal. In order to compare the results between scans (acquired at different time points) it is necessary to apply quantitative or semi-quantitative image analysis methods. True quantitative analysis can only be obtained by performing dynamic imaging combined with the measurement of the arterial input function (IF) in order to apply compartmental models. Such modelling technique is quite complex especially because of the intrinsic difficulties in measuring the IF. It is thus necessary to investigate alternative approaches. One possible choice is to measure the standard uptake value (SUV) defined as the ratio between the tissue radiotracer concentration and the injected activity per animal weight. Firstly in order to obtain accurate SUV values it is necessary to cross calibrate the scanner and to measure the amount of tracer in the animal tail. Estimation of the activity in the tail is quite important considering the intrinsic difficulties in performing good tail injections. Secondly it is necessary to develop a user friendly software tool that take into account all the calibrations including also the uptake time, reconstruction methods, animal weight etc. The objective of this work is to describe the calibration procedures and the software developed at our institution to obtain SUV images of small animals. Material and methods In order to estimate the tracer activity in the animal tail a set of calibrations were carried out using a 0.5 ml syringe (tail phantom) filled with a solution of 18- F ranging from 0.2 to 20 MBq. The coincidences per second (cps) were measured for all the three energy windows (EW) available on our system (GE eXplore Vista) and linear fits between the syringe activity and the total cps were performed. The parameters obtained from the fits (for each EW) were then used to estimate the tail activity from the total cps measured on the tail. The scanner cross calibration was performed by filling a mouse phantom (syringe of 26 mm diameter) with a 18-F concentration ranging from 0.3 to 1 MBq/ml. A set of images were acquired with three energy windows (100-700, 250-700 and 400-700 keV) and recostructed, after Fourier rebinning, using both FBP and OSEM. No corrections for attenuation or scatter were applied. Regions of interest (ROI) were drawn on the reconstructed images and linear fits between the images cps/ml obtained from the ROIs and the known mouse phantom concentrations were performed. All the results obtained using the calibration methods described above were implemented into a graphic user interface (GUI) code developed using IDL 6.2 (Interactive Data Language). The code allows the user to load an interfile image and to obtain as output an interfile “SUV image” that can be loaded into any image processing workstation. In order to evaluate the accuracy of the calibrations several measurements were carried out using three different phantoms. More precisely images of two cylindrical phantoms with different diameters respectively equal to 20 mm and 30 mm covering the entire axial field of view (46 mm) were acquired. The measurements were performed considering 18-F concentrations ranging from 0.3 to 0.9 MBq/ml. The measured radiotracer concentrations were then compared with the known true phantom concentrations in order to calculate the mean error. Secondly images of a custom made cylindrical phantom (with the same dimension as the mouse phantom) consisting in a hot sphere (10 mm diameter) floating into a uniform active background (sphere to background ratio equal to 3) were also acquired. The measured sphere 18-F concentrations were then compared with the true values (ranging from 0,4 to 2 MBq/ml). Results The correlation coefficients of the linear fits obtained using the tail phantom, mouse phantom and hot sphere phantom were always greater than 0.999, showing a good correlation between the measured cps/ml and the known true 18-F concentration. For the 100-700 keV EW the mean differences between the measured and true phantoms concentrations were respectively equal to 3.5 % and 2.2% for FBP and OSEM. The mean differences between the concentrations for the 250-700 keV EW were respectively equal to 5.5% and 4.7% for FBP and OSEM. For the 400-700 keV EW the mean concentrations differences were respectively equal to 5.4% and 4.5% for FBP and OSEM. Results obtained using the hot sphere phantom show that for FBP the mean differences between the measured and true hot sphere concentrations were respectively equal to: 12%, 7% and 2% for the 100-700 keV, 250-700 keV and 400-700 keV EW. For OSEM reconstruction the mean concentrations differences were equal to: 8%, 5% and 3% for the 100-700 keV, 250-700 keV and 400-700 keV EW respectively. Conclusions Results show that it is possible to obtain reasonably accurate calibrations and, thus, to obtain good estimate of radiotracer concentration in small animals. The mean errors were slightly higher when using the hot sphere phantom compared with the cylindrical phantoms, however in this case a much larger range of 18-F concentrations was considered. The developed code was also tested by several users showing good stability.