Comparison and Optimization of Fitting Methods for the Quantification of Multi-Parametric Relaxometric GE-SE EPIK Data

Motivation: Imperfect refocusing pulses and imaging noise hinder the quantification performance of simultaneous T2 and T2* fitting approaches and require specific correction methods. Goal(s): Improve fitting results using a combination of slice profile correction terms and sophisticated fitting models to consider noise. Approach: Individual free fitting parameters are added for signals acquired following the two refocusing pulses, and different fitting models are compared: NLSQ, offset-noise model and Rician distributions of first and second order. Results: Correction parameters significantly improve the resulting T2 and T2* distributions. Offset-model fitting underestimates relaxation times, while Rician models improve the fit precision and significantly reduce residuums. Impact: Individual free pulse correction parameters for subsequent refocusing pulses combined with noise consideration in the form of Rician model fitting significantly improve simultaneous T2 and T2* quantification based on GE-SE EPIK data in terms of better precision and reduced residuums.