Moving slice septa and pseudo three-dimensional reconstruction for multi-ring PET

Abstract

Describes moving slice septa for reducing scattered and random events in multi-ring positron emission tomographs (PET) and a new simple method of three-dimensional (3D) image reconstruction which is useful when the maximum axial acceptance angle of coincidence is not too large. The moving septa considered are linear and sinusoidal piston motions of parallel septa, axis wobbling of parallel septa and rotation of spiral septa. The proposed reconstruction algorithm is basically a filtered (1D) backprojection (3D) method. The low frequency image is reconstructed using direct (and cross) plane coincidence as in the conventional way, and the high frequency image is reconstructed using all projection data through high-pass filtering. The two images are superimposed in such a way that the final frequency response is normal. The axial cross talk is effectively eliminated by Gaussian high-pass filtering with negligible increase in statistical noise, and it is possible to include most of the oblique coincidence events in the reconstruction in a single pass. Simulation studies with a maximum axial acceptance angle of +or-7.6 degrees (+or-10 ring difference) show satisfactory results.