A dynamic expectation maximization algorithm for single camera rotation dynamic SPECT (dSPECT)

A new dynamic SPECT reconstruction algorithm is presented that is able to estimate the dynamic behaviour of activity within an object from projection data acquired with a conventional single slow rotation of a SPECT camera. The algorithm is based on the static expectation maximization (EM) reconstruction algorithm, but incorporates dynamic inequality constraints in the reconstructed object voxel activity over time to account for dynamic behaviour. This method has previously been presented in the constrained least squares (CLS) dynamic SPECT method which uses an optimization technique to minimize a dynamic objective function. A series of simulated dynamic object projection data sets have been reconstructed with the new dynamic EM algorithm (dEM) and results compared to the CLS method of dynamic image reconstruction. Results of these tests indicate that the dEM algorithm is able to produce time activity curve parameters that are within 20% of the true values, while improving upon reconstruction time by a factor of 2-3 times compared to the CLS method. Further experiments have been performed using the authors' dynamic heart-in-thorax phantom and have implemented both attenuation and depth dependent detector response corrections. Results indicate a similar degree of accuracy when reconstructed with the dEM algorithm, with again an improvement in reconstruction time of between 2-3 times.