Wavelet Thresholding-Based Denoising Method of List-Mode MLEM Algorithm for Compton Imaging

Iterative image reconstruction of data measured by a Compton scattering camera has to overcome various difficulties, e.g., large amount of data, noise arising from both low counts recorded, and the imaging response. Image estimation by the Maximum Likelihood (ML) criterion induces noise amplification, so a denoising step is needed. The proposed solution is a denoising technique using wavelet-based thresholding of the ML Expectation-Maximization (EM) update factors, called WTDEM. The thresholds are scale dependent, and proportional to the standard deviation of the high-frequency sub-band coefficients at the respective scale. It results in lower reconstruction errors than the MLEM algorithm, and the Gaussian smoothing, and, in addition, it is stable. The WTDEM algorithm is illustrated by computer experiments.