On quantification errors of $$ {R}_2^{\ast } $$ and proton density fat fraction mapping in trabecularized bone marrow in the static dephasing regime
Sophia Kronthaler, Maximilian N. Diefenbach, Christof Boehm, Mark Zamskiy, Marcus R. Makowski, Thomas Baum, Nico Sollmann, Dimitrios C. Karampinos
Abstract
Purpose
To study the effect of field inhomogeneity distributions in trabecularized bone regions on the gradient echo (GRE) signal with short TEs and to characterize quantification errors on R2* and proton density fat fraction (PDFF) maps when using a water-fat model with an exponential R2* decay model at short TEs.
Methods
Field distortions were simulated based on a trabecular bone micro CT dataset. Simulations were performed for different bone volume fractions (BV/TV) and for different bone-fat composition values. A multi-TE UTE acquisition was developed to acquire multiple UTEs with random order to minimize eddy currents. The acquisition was validated in phantoms and applied in vivo in a volunteer’s ankle and knee. Chemical shift encoded MRI (CSE-MRI) based on a Cartesian multi-TE GRE scan was acquired in the spine of patients with metastatic bone disease.
Results
Simulations showed that signal deviations from the exponential signal decay at short TEs were more prominent for a higher BV/TV. UTE multi-TE measurements reproduced in vivo the simulation-based predicted behavior. In regions with high BV/TV, the presence of field inhomogeneities induced an R2* underestimation in trabecularized bone marrow when using CSE-MRI at 3T with a short TE.
Conclusion
R2* can be underestimated when using short TEs (<2 ms at 3 T) and a water-fat model with an exponential R2* decay model in multi-echo GRE acquisitions of trabecularized bone marrow.