Comparison of two fast reconstruction methods for a cylindrical PET camera

The authors compare the accuracy and precision of two fast reconstruction methods for a cylindrical PET (positron emission tomography) camera. One method is based on the estimation of the missing data; the other uses different amounts of data to reconstruct each slice. Both methods are implemented in the frequency domain and run on realistic data in less than two hours. Their intrinsic quality is quite similar. In the presence of statistical noise, there is no significant difference between the methods. One could therefore prefer to use the interpolative reconstruction for a small centered activity distribution like a receptor study, while reserving the Favor procedure for a large distribution pattern like a cerebral blood flow study. This study has been carried out on the model of the cylindrical PET camera under development at the Massachusetts General Hospital.<>