Characterization of Multimodal Preclinical Imaging Systems

Sammanfattning

BackgroundA setup of four compatible imaging systems for small animal from Bruker has been delivered toSahlgrenska Academy at University of Gothenburg and Sahlgrenska University Hospital, and forms thebasis of SBIC (Sahlgrenska Bioimaging Center). The aim of the study was to characterize theperformance of the positron emission tomography (PET)system and computed tomography (CT)systemat SBIC as well as the dose calibrator and two radiation protection instruments, but also to analyse twomethods of attenuation correction using either data from CT or magnetic resonance imaging (MRI).MethodThe response of the dose calibrator was characterized for geometric factors and activity dependenceusing 99mTc, 177Lu, 131I, 125I and 18F sources. The PET-system was characterized by measuring thesensitivity, correction factors, linearity, highest and lowest measurable activity, scatter fraction, spatialresolution, uniformity and recovery coefficients using various phantoms filled with 18F-FDG(fluorodeoxyglucose). To characterize the CT-system three protocols, two reconstruction algorithms,uniformity, spatial resolution and noise were measured using two phantoms. CT- and MRI-basedattenuation correction of PET images of a tumour bearing mouse injected with 68Ga-DOTATOC wasanalysed by comparing the activity concentrations in various tissues. Two radiation protectioninstruments, Berthold LB 122 and 124, were characterized by examining the linearity and minimumdetectable activity for two phantoms containing 131I. ResultsUsing the dose calibrator, a linearity was identified for all radionuclides, the suggested working regionwas 2-8 cm, larger uncertainties were found for smaller volumes, and a radionuclide dependence wasidentified. The sensitivity of the PET-system varied with activity, however, a linear correlation wasidentified with enlarged uncertainties for lower activities. Best uniformity and resolution were achievedfrom the image without attenuation correction. Highest and lowest measurable activity was 41 MBq and0.1 MBq, respectively. The CT-system achieved the lowest variation of Hounsfield units (HU) whenusing the high-resolution protocol, but the resolution was similar for the high-resolution and the generalpurpose protocols. The CT-attenuation correction of PET images resulted in higher activityconcentration than MRI-attenuation correction. A linear response was identified for both radiationprotection instruments for each measurement setup. The minimum detectable activity was estimated to35 kBq at 10 cm distance using Berthold LB 122.ConclusionsThe dose calibrator is appropriate to use under the suggested working conditions.Essay/Thesis: 30 hpProgram and/or course: Medical physicsLevel: Second CycleTerm/year: Fall 2025Supervisor:Eva Forssell-Aronsson, Hana Bakr, Johan Spetz, Lukas Lundholm,Mikael MonteliusExaminer: Magnus BåthKeywords:Preclinical imaging, Multimodal, Characterization, small animal PET,micro-CT, MRI attenuation correction, Bruker, Molecubes3The sensitivity of the PET-system, calculated in the software PMOD, was activity dependent. Thesystem had a resolution of at least down to 2 mm and the measurable 18F-FDG activity range was 0.1-37 MBq. For the CT-system, the Feldkamp, Davis and Kress (FDK) algorithm with helical acquisitionwas less good and the image space reconstruction algorithm (ISRA) should be usen when quantificationis needed. The MRI-based attenuation correction method evaluated underestimated the activityconcentration compared with CT-based attenuation correction and needs to be improved. Altogether,the PET- and CT-systems were well characterized, but further studies on phantoms and animals wouldbe interesting and give more information.The two radiation protection instruments worked well.