Frequency domain method for computing missing input data for a cylindrical PET camera

Input projection data sets at nonzero tilt angles to the transverse planes may miss seeing activity at the ends of the cylindrical volume of a cylindrical positron emission tomographic (PET) camera. To use the data at these tile angles, it is necessary to compute and fill in with estimates of the missing parts these data. In principle, one could make an initial reconstruction using zero-tilt-angle data, and then compute the missing data by reprojecting the initial reconstruction along the desired rays. Computer simulation results are presented of a method that computes the missing data in the frequency domain and thereby eliminates the computationally long process of forward projection. In these initial computer simulation tests the method for computing missing data worked well, producing in the reconstructed image principally a magnification of error that would have existed with no missing data.<>