Estimation of signal and noise from Rician distributed data
Jan Sijbers, Arnold Jan den Dekker, Dirk Van Dyck, Erik R. Raman · 1998
Conventional estimation methods applied to Rician distributed data, (such as magnitude magnetic resonance data) yield biased results. In our work, it is shown where the bias appears. Furthermore, a novel estimation technique, based on Maximum Likelihood estimation, is developed for optimal estimation of signal as well as noise from Rician distributed data. It is shown that the proposed method is superior in terms of mean squared error compared to the performance of conventional estimation techniques.