List-event data resampling for quantitative improvement of small animal Yttrium-90 PET image

Objectives Positron emission tomography (PET) imaging of yttrium-90 (Y-90) has been received much interest especially since after Y-90 used as treatment for liver tumor. The purpose of present study was improving the image quality and evaluating the noise and statistical data of PET imaging of Y-90 by using non-parametric bootstrap resampling.

Methods Inveon PET list-event data was acquired as static data for 20 min after 60 min of Y-90 injection via mouse tail vein. List-event data format consist of packet of 48bit, 8 bit header and 40 bit payload space. The first bit is always zero and the next three bit is gray-code complementing the problems of binary code and also evaluating sequence and flow of each packet. Resampled event data of original list-event was changed to realigned sinogram by using non-parametric bootstrap. Realigned sinogram was reconstructed for imaging by using OSEM 2D algorithm with 16 subset and 4 iterations. Regions of interest (ROI) was drawn around liver as the target and background. The contrast, uniformity and signal to noise ratio (SNR) of the reconstructed image were then calculated.

Results As a result, contrast, uniformity and SNR of four cases, original, bootstrapped samples of 2, 10 and 300, were obtained. The contrast values of original, bootstrapped samples of 2, 10 and 300 were 1.85, 1.12, 2.12, 2.27 and uniformity values were 2.22, 2.12, 1.75, 1.85, respectively. Both results of contrast and uniformity showed that the more number of bootstrapped samples get better result. In case of SNR, values were 86.28, 61.71 (28.46% decreased), 107.69 (24.81% increased), 111.81 (29.59% increased), respectively.

Conclusions The results informed that non-parametric bootstrap resampling technique provides the PET image whose contrast, uniformity and SNR were improved and it also suggested that quantitative Y-90 PET image is acceptable measures of diagnosis and therapeutic effects.