JAVA-based plugin for tomographic reconstruction in SPECT

The aim of this study was to develop a plugin for software ImageJ able to do tomographic reconstruction for SPECT. Two methods of tomographic reconstruction for emission images were implemented: filtered backprojection (FBP) and an maximum likelihood-expectation maximization (MLEM). Filters were also implemented. Validation tests were carried out with simulated images of geometric models and an anthropomorphic model through Monte Carlo method. The analysis of convergence of the iterative method through the estimate of the relationship between the average uptake and the standard deviation resulted in an indication of reconstruction with 10 to 15 iterations. The images reconstructed through the ML-EM method with attenuation correction presented better visual quality than the non-corrected images. For 128x128 pixel images, MLEM time is 33 times longer than FBP. When attenuation correction is carried out time increases 3.5x.