A Reconstruction Method Based on Deep Convolutional Neural Network for SPECT Imaging

We explore a novel method for tomographic image reconstruction in the field of Single Photon Emission Computerized Tomography (SPECT) imaging. Deep Learning methodologies and more specifically deep convolutional neural networks (CNN) are employed in the new reconstruction method, which is referred to as "CNN Reconstruction - CNNR". Projection data were obtained from software phantoms and were used to evaluate the efficacy of the method. The resulting tomographic images are compared to those produced by the "Maximum Likelihood Expectation Maximization" (MLEM) and the "Filtered Back Projection" (FBP). The results show that the proposed CNNR model outperforms both FBP and MLEM methods.