TV Based Low Dose PET Image Reconstruction

Positron Emission Tomography (PET) uses the radioactive tracer injections. The use of radioactive tracer needed in PET detection can make the radiation personnel health risk. Some research show that:with close contact radioactive nuclide, chances of being radiation personnel cancer will be much higher than the normal people. And the cost of radioactive tracer occupies a certain proportion of PET. Thus the reduction of injection dose in PET is necessary.Low dose PET reconstruction is a challenging issue for PET reconstruction methods. Because the measurement data is with low signal-to-noise ratio(SNR). PET image reconstruction is an ill-conditioning inversion problem, so proper prior knowledge should be incorporated to constrain the reconstruction. The work of this paper can be summarized as:(1) Since PET images have the character of piecewise smoothing, we propose the total variational (TV) minimization based algorithm for low dose PET imaging. Due to the power of TV regularization, this reconstruction strategy is able to preserve the small features while keeping the edge information clear. To solve the objective function, the alternative minimization method has been employed with improved computational speed and robustness.(2) Considering of the statistic characteristics of observation of PET system, we propose the Poisson-TV method with a Poisson data fitting term in the objective function. The objective function is solved by a iterative method. Experimental results on Monte Carlo simulations, physical phantom and real patient data demonstrate its superior performance.(3) We apply Monte Carlo simulation to generate the PET measurement data with different counting rate. These data are used to validate the effect of proposed methods in low dose PET reconstruction in this paper. This work provides a possible approach to realize the low dose PET.