Alternating direction method of multipliers applied to medical image restoration

We investigate the effects of the regularization parameter for the norm () and penalty parameter () in the alternating direction method of multipliers (ADMM) on the quality of restored medical images. Simulation studies are performed using images degraded by a point spread function (PSF) and Gaussian noise. The j-th column of the system matrix () is calculated by convolving the image with unity at pixel j and zero at all other pixels and the PSF. The simulation studies show that the mean structural similarity index is maximal when is approximately 10 to 20, where , with and being the transpose of A and the observed data, respectively. The restored image became blurred with a decrease in . This study will be useful for identifying optimal parameter values in the ADMM when applied to medical image restoration.