The use of the least-squares minimal residual algorithm for fast and regularized attenuation compensation in SPECT

This paper addresses the problem of attenuation compensation in SPECT imaging. It describes a fast and regularized iterative algorithm for reconstructing cardiac SPECT images acquired with a 180° protocol and in the presence of nonuniform attenuation. The proposed reconstruction method is based on the use of the least-squares minimal residual minimization algorithm. Since the inverse problem is ill-posed, regularization must be applied in order to get a stable reconstruction. The authors present a spatially adaptive regularization technique. Results obtained with numerical simulations, anatomic phantoms and clinical data show that the proposed reconstruction method is two to four times faster than methods based on preconditioned conjugate gradient. They also demonstrate the effectiveness of the spatially adaptive regularization technique