Edematous alterations and fatty infiltration of skeletal muscles are main characteristics of neuromuscular diseases. Both pathologies increase the total muscle T2 relaxation-time^1-3. T2 has been routinely quantified by multi-echo spin-echo (MESE) with known problems related to the refocusing pulses in presence of B1-inhomogeneity (3D-MESE) and additional slice profiles issues (2D-MESE)⁴. An adiabatic BIR-4 RF pulse previously used for generating T2-weighted contrast⁵ and recently adjusted with gaps (modified BIR-4)⁶ can be alternatively used for B1-insensitive T2 mapping. In addition, presence of fat within muscle can influence T2 quantification⁷. Purpose of this work was to compare a newly developed T2 mapping-sequence (T2Prep-3DTSE) with 2D- and 3D-MESE regarding T2 values, influence of B1-field and proton density fat fraction (PDFF) on it as well as reproducibility in thigh muscles of healthy subjects.

Ten young and healthy subjects (age: 27.2±1.7yrs, BMI: 22.9±4.3kg/m², n=5 male) with no history of diabetes, neuromuscular disorders and Quadriceps muscle injuries were recruited.

MR measurements: The bilateral thigh muscles were scanned on a 3.0T system (Ingenia, Philips Healthcare) using anterior and posterior coil arrays. The FOV was centered at the femur mid-length using greater trochanter and tibial plateau as anatomical landmarks⁷. T2 mapping was performed using 2D-MESE, 3D-MESE and T2prep-3DTSE. Three subjects were scanned three times to assess the reproducibility of the three methods. The T2prep-3DTSE used T2 preparation based on a modified BIR-4 RF pulse, where two gaps with equal duration were introduced to achieve a module with variable TE (Fig. 1). For the MESE-sequences standard implementations were used. SPAIR was used for all sequences. The common sequence parameters were: FOV = 42×26×12cm³, TR/TE = 1.5 s/20ms and acquisition voxel = 2×2×4mm³. The TSE sequence had a TSE factor = 50 and T2Prep durations of 20/40/60/80ms. In MESE sequences echoes have been acquired in 10ms steps. Additionally, B1-map was acquired with dual TR method, B0- and PDFF-maps were measured based on a 6-echo gradient echo sequence.

Data analysis: Data at the same echo times for all three T2 mapping sequences was fitted by a 2-parameter fit. The first echo was excluded for MESE data. B0 and PDFF maps were determined based on a signal model accounting for the multi-peak fat spectrum and single T2* decay effects. Two muscles (Rectus femoris, Vastus lateralis) of the Quadriceps muscle, mainly affected in neuromuscular disorders, were selected. ROIs were drawn manually in the interior of these muscles avoiding vessels and fasciae. To assess reproducibility errors of the different T2 mapping-sequences, root mean square coefficients of variation (RMSCV)⁸ were computed. T2 values of the different T2 mapping sequences were compared and correlations were calculated between T2 values and PDFF and B1.

Representative T2 maps and the respective B1-map are shown in Fig. 2. When the B1 error was large, T2 was significantly higher with the MESE-sequences compared to T2Prep-3DTSE (arrow in Fig. 2). When the B1 error was small, T2 values from 2D-MESE were higher than T2 values from 3D-MESE and T2Prep-3DTSE (circle in Fig. 2). Significant differences between T2 values of T2Prep-3DTSE and T2 values of 2D-MESE respectively 3D-MESE were found in all four muscles with the lowest values for T2Prep-3DTSE (p<0.05; Fig. 3). No significant correlations were found between PDFF and T2 values. The left Rectus femoris muscle showed the highest variance of T2 values (T2Prep-3DTSE: 30.94±1.28ms, 3D-MESE: 41.7±7.8ms; 2D-MESE: 36.6±2.4ms) and B1-field (54.1±15.6%) with RMSCVs: T2Prep-3DTSE: 1.7%; 3D-MESE: 19.9%; 2D-MESE: 1.8% (Fig. 4). Significant negative correlations between T2values and B1-field were found in the left Rectus femoris muscle for 3D-MESE (r=-0.86, p<0.001) and 2D-MESE (r=-0.54, p=0.033), but not for T2Prep-3DTSE (p=0.71; Fig. 5). The results show a low reproducibility of the T2 values measured by 3D-MESE. Both, T2 values of 3D-MESE and 2D-MESE are strongly affected by an inhomogeneous B1 field. MESEs suffer from the presence of stimulated echoes, induced by B1 effects (2D- or 3D-MESE) and slice profile effects (2D-MESE). Recent work has shown that the application of extended phase graphs can compensate for stimulated echoes due to B1 effects in MESE-based T2 quantification⁹. However, MESE can lead to T2 overestimation in fatty infiltrated muscles (see subcutaneous fat region; Fig. 2), whereas T2prep-3DTSE leads to T2 values in subcutaneous fat in the same range as healthy muscle (Fig. 2). The remaining fat signal after SPAIR in the three sequences requires further investigations. In conclusion, we proposed the use of a B1-insensitive T2-prepared 3D TSE for robust muscle T2 mapping and showed that the proposed method gave reliable and reproducible T2 values in the thigh muscles of ten healthy volunteers.