Skip to Main Content
In a previous work, we showed that anatomical priors can improve lesion detection in simulated Ga67 images of the chest. We herein expand and enhance our previous investigations by adding scatter in the projections, by using the triple energy window scatter compensation method and by implementing a new scheme for image reconstruction. Phantom images are created using the SIMIND Monte Carlo simulation software and the mathematical cardiac-torso (MCAT) phantom. The anatomical data are the original, noise-free slices of the MCAT phantom. Images are reconstructed using the DePierro algorithm. Two weights for the prior are tested (0.005 and 0.02). The following reconstruction scheme is used to reach convergence: The 120 projections are reconstructed successively with 4, 8, 24, 60, and 120 projections per subset with 1,1,1,1, and 50 iterations respectively; the result of each reconstruction is used as an initial estimate for the next reconstruction. Several strategies were investigated: no anatomical prior information, and anatomical information for organs and/or lesion. Lesion detection was performed by a numerical observer with an LROC task. Strategies including anatomical priors yield better results in terms of lesion detection, as compared to the strategy using no prior and only post-reconstruction Gaussian smoothing.