PT  - JOURNAL ARTICLE
AU  - Wei Chen
AU  - Ying Liang
TI  - Impact of BSREM reconstruction algorithm on image quality and quantification accuracy of PET-a phantom study
DP  - 2021 May 01
TA  - Journal of Nuclear Medicine
PG  - 3036--3036
VI  - 62
IP  - supplement 1
4099  - http://jnm.snmjournals.org/content/62/supplement_1/3036.short
4100  - http://jnm.snmjournals.org/content/62/supplement_1/3036.full
SO  - J Nucl Med2021 May 01; 62
AB  - 3036Objectives: We performed this phantom study to evaluate the impact of BlockSequential Regularized Expectation Maximization (BSREM) reconstruction algorithmwith different beta values, which control the relative strength of the regularization, onimage quality and quantification accuracy of PET. Methods: We filled F-18 solution in the body of a NEMA/IEC image quality phantomwith a concentration of 5.3 kBq/ml. The F-18 solution concentration ratio in spheresand background was 4:1. PET image was acquired on a 3 ring DMI from GE Healthcarewith 2 minutes/bed. BSREM algorithm was used to reconstruct the PET image.Attenuation correction was done with low dose CT. 3D-ROI was delineated on sphereand background based upon a threshold of 42% maximum activity. Coefficient ofvariance (CV), target-to-background ratio (TBR), contrast noise ratio (CNR), hotsphere recovery coefficient (HRC), and volume recovery coefficient (VRC) in eachsphere were assessed. Results: TBR and CNR increased along with increased beta value, while CV was inthe opposite except the smallest sphere (10mm-diameter). For the smallest hot sphere,TBR reached a plateau after the beta value arrived at 300, while CNR declined slowlyafterward. VRC increased along with increased beta value, while HRC was in theopposite. Conclusions: Beta value 300 seems a favorable choice in fluorine-based PET scan withBSREM reconstruction in consideration of image quality and lesion detectability. Ahigh beta value tends to bring high volume recovery, but a low activity recovery. Keywords: Positron Emission Computed Tomography,Block Sequential RegularizedExpectation Maximization,Image quality,Quantitative analysis