Maximum a posteriori estimation for SPECT using regularization techniques on massively-parallel computers

Single photon emission computed tomography (SPECT) reconstructions were performed using maximum a posteriori (penalized likelihood) estimation via the expectation maximization algorithm. Due to the large number of computations, the algorithms were performed on a massively parallel single-instruction multiple-data computer. Computation times for 200 iterations using Good´s rotationally invariant roughness penalty were on the order of 5 min for a 64*64 image with 96 view angles on a 4096 processor machine, and 40 s on a 16384 processor machine. Computer simulations were performed using parameters for the Siemens gamma camera and clinical brain scan parameters comparing two regularization techniques to conventional reconstructions. Regularization by kernel sieves and penalized likelihood with Good´s rotationally invariant roughness measure are compared to filtered backprojection. Twenty-five independent sets of data are reconstructed for the pie and Hoffman brain phantoms.<>