Improved Quality of 3-D Compton Images by List-Mode EMTV Reconstruction

The Compton camera does not require a collimator and can detect gamma rays in a wide energy range, so it is expected as a new nuclear medicine modality. The aim of this study is to improve the quality of three-dimensional (3-D) Compton images by reconstruction using total variation (TV) regularization. We incorporated TV regularization into the list-mode maximum likelihood expectation maximization (LM-MLEM) reconstruction method for 3-D Compton imaging. This is called list-mode expectation maximization total variation (LM-EMTV). A Monte Carlo simulation of a Compton camera was performed with a Cs-137 uniform plane source, projection data in 12 directions were acquired, and the image quality was evaluated by the uniformity and the edge blurring of the reconstructed image. We also examined the possibility of reducing the number of projections. From results of simulation, the uniformity of image reconstructed using LM-EMTV improved 66% without edge degradation. In addition, even if the number of projections reduced to directions, which is half of the 12 directions, an image with the same quality as that by the conventional MLEM method was obtained. In conclusion, LM-EMTV image reconstruction method improved the quality of the 3-D Compton images, suggesting the possibility of reducing the number of projections.