Muscle water T2 has been proposed as an imaging biomarker of disease activity in neuromuscular diseases. 2D multi-echo spin-echo sequences have been used for muscle T2 mapping with known limitations including the sensitivity to transmit B1 inhomogeneities. Confounding effects on the T2 quantification can be removed by matching the experimental signal with pre-simulated theoretical signal decay curves obtained by Bloch simulations. However, up to now it is unclear what discretization over the slice profile is sufficient to determine a meaningful dictionary. The purpose of this work was to utilize the configuration model in order to determine a minimal number of z-location necessary for the simulation and applies the technique for T2 mapping of in vivo thigh musculature data.