In this work we present an evaluation approach based on artificial neural networks (ANN) for fitting the IVIM-Kurtosis model parameters on the basis of simulated DWI data. The ANN approach is compared to an ordinary bounded least squares regression (LSR) in terms of correlation between estimates and ground truth, systematic, statistical and total estimation error. While for D and K high correlations and low errors were found for both LSR and ANN, a significant improvement was observed for f and D* regarding correlation coefficients, precision and the total estimation error when using ANN.
This abstract and the presentation materials are available to members only; a login is required.