Optimisation of scatter correction based on the use of partially transparent beam stoppers in PET

Abstract

To improve the counting efficiency, modern positron emission tomography scanners generally perform 3D data acquisition with the septa retracted. As a consequence, the scatter fraction can surge to as high as 40%. Our previous method proposed the use of a beam stoppers (BS) device to directly measure the scatter and to remove it from the sinogram. The BS partially attenuate the primary radiation of certain lines of response (LOR) without affecting the scattering events. To measure the scatter, dual scans are performed with and without the device in place. This study proposes an improved method. The dual scans are performed with the device placed at two different locations. The proposed method reduces the effective sampling distance without the need to actually increase the stopper's number. The scatter component at the LOR blocked by each stopper can be estimated. Assuming that the scattered radiation has a spatially slow-varying distribution, the whole scattered sinogram can be recovered using interpolation from these local measurements. Once the scatter component is estimated, the sinogram of the primary radiation can be obtained by subtracting the scatter component from the original data. The scatter free image is then reconstructed from the primary sinogram. This method was applied to scanned data of two digital phantoms using the Simulation System for Emission Tomography simulation package to demonstrate its effectiveness. Preliminary results showed that this method has slightly better performance than the previous BS method and the conventional single scatter simulation technique with smaller mean squared error and less noise in the corrected image.