Markus Nilsson1, Erik Alerstam2, Sara Brockstedt3, Ronnie Wirestam1, Freddy Sthlberg1,4, Jimmy Ltt3
1Department of Medical Radiation Physics, Lund University, Lund, Sweden; 2Department of Physics, Lund University, Lund, Sweden; 3MR Department, Center for Medical Imaging and Physiology, Lund University Hospital, Lund, Sweden; 4Department of Diagnostic Radiology, Lund University, Lund, Sweden
Evaluation of diffusion MRI data obtained with different diffusion times currently involves non-linear fitting of analytical models. These models rely on assumptions about the tissue as well as on mathematical approximations in the derivation of the signal expression. Replacing the non-linear fitting by a lookup database, constructed from fast Monte Carlo simulations, improves and speeds up the evaluations. This novel approach was demonstrated by investigating the estimated posterior distributions for four tissue parameters, i.e. the intracellular volume fraction, the cell diameter, the intracellular exchange time and the apparent diffusion coefficient, given simulated signal-versus-b curves with added noise.