Evaluation of an iterative cascade gamma ray correction algorithm for non-standard PET nuclides at various counting statistics in high resolution small animal PET imaging

We describe an evaluation of an iterative cascade gamma ray correction algorithm which was developed to improve the quantitative accuracy of small animal PET imaging with non-standard PET nuclides. The cascade correction algorithm uses the emission image and the attenuation map of the object to compute the shape or the spatial distribution of the coincidences caused by the cascade gamma rays, and the distribution is then scaled to the measured coincidences similarly to the typical scatter estimation methods. The sinogram profiles of the cascade estimate were compared with those of the scatter estimate and the prompts - randoms, and the cascade fraction was evaluated at various counting statistics for various random fractions and count rates. A quick convergence for the shape and the magnitude of the cascade estimate was observed and found to be not sensitive to the number of counts or count rate within the frame. A higher variation of the cascade fraction was observed for frames with lower number of counts. In addition, a higher overestimation of the cascade fraction was observed for frames with a higher random fraction and/or with a lower number of counts. As a result, a calibration for the scaling of the cascade estimate was applied for the high random fraction and low count situations, and the improvement was demonstrated using the sinogram profiles.