To develop a dedicated algorithm allowing to quantify, from a single MR acquisition, fat and muscle fractions together with magnetic susceptibility and transverse relaxation time (T₂*) and to assess dystrophic patients on the basis of specific quantitative indices.

Patients: MR acquisition was performed at 1.5T using a 3D multiple-echo spoiled gradient echo sequence in 9 patients with a facio-scapulo-humeral dystrophy and 8 controls. According to commonly used clinical and functional scales, two subgroups of patients were distinguished (moderate (MOD) and severe (SEV)).

Image reconstruction: Phase images were unwrapped to compute the B₀ field inhomogeneities (ΔB₀) map and the ΔB₀-demodulated real part images. The real part was used for fat-water separation as previously described (1). The fat-water separation algorithm provided T₂* and PDFF maps. From the ΔB₀ map, external (B_out) and internal (B_int) fields were separated with the projection into dipole field (2). From B_int, the dipole inversion was performed with a single orientation Bayesian regularization including spatial priors derived from magnitude images to compute the susceptibility map (3). A segmentation algorithm (4) was used to calculate the index of fatty infiltration (IFI) representing the relative amount of fatty-infiltrated pixels and defines the normal appearing muscle area from which the distributions of PDFF, T₂* and magnetic susceptibility were analyzed (Figure 1).

The SEV group displayed a significantly larger fatty infiltration (p<0.01) of the normal-appearing muscle compartment i.e. PDFF_muscle = 22.0 ± 4.7% as compared to the control group (11.3 ± 3.7%) whereas no significant difference was found between the MOD (13.8 ± 2.1%) and the control group. In the normal-appearing muscular compartment, the averaged T₂* value was significantly higher (p<0.05) in both the MOD (30.4 ± 0.8 ms) and the SEV (30.4 ± 1.7 ms) groups as compared to the controls (29.2 ± 0.5 ms). No significant difference was found between the patients groups. T₂* FWHM (Full Width at Half Maximum) significantly increased (p<0.01) with respect to the groups (10.6 ± 1.0, 12.2 ± 1.4 and 17.6 ± 3.3 ms in the control, MOD and SEV group respectively). The averaged magnetic susceptibility significantly decreased with respect to the groups: -0.89 ± 0.29; -1.49 ± 0.34 and -2.06 ± 0.16 ppm in the control, MOD and SEV group respectively whereas magnetic susceptibility FWHM increased: 7.2 ± 1.4, 8.2 ± 0.4 and 9.4 ± 0.5 ppm in the control, MOD and SEV group respectively).The present results illustrated that QSM could be performed in skeletal muscle at 1.5T and that the corresponding algorithm can be included in the fat-water separation pipeline. The dedicated segmentation method allowed us to quantify the corresponding indices in the normal-appearing muscle compartment. The short echo-time spacing, mandatory for an accurate fat-water separation, is advantageous for the QSM reconstruction given that wraps between echoes can be consistently reduced thereby simplifying the unwrapping procedure. However, the phase-to-noise ratio was not optimal since the echo train length remains shorter than T₂*. Nevertheless, this effect was limited by the high number of echoes. On that basis, it has been possible to obtain a multiparametric information from a single MR acquisition. While magnetic susceptibility-related information can be added to the relaxometry and fat fraction measurements for the assessement of disease severity and progression, further works would be mandatory in order to delineate the exact mechanisms leading to magnetic susceptibility changes in the normal appearing muscle and more particularly to understand the exact links with edema, inflammation, fibrosis and macromolecule deposition.