Low dose PET reconstruction with total variation regularization

Low dose positron emission tomography(PET) reconstruction remains a challenging issue for statistical PET reconstruction methods due to the low SNR of data. Due to the ill-conditioning of image reconstruction, proper prior knowledge should be incorporated to constrain the reconstruction. Since PET images are piecewise smoothing, we propose the total variational (TV) minimization based algorithm for low dose PET imaging. The fundamental power of this strategy rests with the edge locations of important image features tend to be preserved thanks to TV regularization. In addition, a new computational method have been employed with improved computational speed and robustness. Experimental results on Monte Carlo simulations demonstrate its superior performance.