Tehran University of Medical Sciences
home | Farsi

Davood Beiki

The influence of resolusion recovery by using collimator detector response during 3D OSEM image reconstruction on 99mTc-ECD brain SPET images

Authors: Kalantari F, Rajabi H, Ay MR, Razavi-Ratki SK, Fard-Esfahani A, Beiki D, Eftekhari M, Fallahi B, Sadeghian L, Emami-Ardekani A,
Keywords: Brain 99mTc-ECD SPET, Image resolution, Image reconstruction, Iterations
Hell J Nucl Med , Vol.15, No.2, 2012,Page:92-97

Partial volume effect, due to the poor spatial resolution of single photon emission tomography (SPET), significantly restricts the absolute quantification of the regional brain uptake and limits the accuracy of the absolute measurement of blood flow. In this study the importance of compensation for the collimator-detector response (CDR) in the technetium-99m ethyl cysteinate dimer (99mTc-ECD) brain SPET was assessed, by incorporating system response in the ordered-subsets expectation maximization (OSEM) reconstruction algorithm. By placing a point source of 99mTc at different distances from the face of the collimator, CDR were found and modeled using Gaussian functions. A fillable slice of the brain phantom was designed and filled by 99mTc. Projections acquired from the phantom and also 4 patients who underwent the 99mTc-ECD brain SPET were used in this study. To reconstruct the images, 3D OSEM algorithm was used. System blurring functions were modeled, during the reconstruction in both projection and backprojection steps. Our results were compared with the conventional resolution recovery using Metz filter in filtered backprojection (FBP). Visual inspection of the images was performed by six nuclear medicine specialists. Quantitative analysis was also studied by calculating the contrast and the count density of the reconstructed images. For the phantom images, background counts and noise were decreased by 3D OSEM compared to the FBP-Metz method. Quantitatively, the ratio of the counts of the occupied hot region to that of the cold region of the reconstructed by FBP-Metz images was 1.14. This value was decreased from 1.12 to 0.86 for 3D OSEM of 2 and 30 iterations respectively. The reference value was 0.85 for the planar image. For clinical images, hot to cold regions (grey to white matter), the count ratio was increased from 1.44 in FBP-Metz to 3.2 and 4 in 3D OSEM with 10 and 20 iterations respectively. Based on the interpretability of images, the best scores (3.79±0.51