Facioscapulohumeral muscular dystrophy is characterized by a peculiar non-linear muscle-by-muscle involvement and is very hard to predict. The purpose of this study was to use dynamic phase contrast MR imaging of electrically stimulated quadriceps muscle to characterize the elastic potential of the muscle in FSHD patients and contribute to the understanding of this challenging disease. Velocity, strain, and strain rate were analyzed and compared to the results of physical examination.
18 patients (13 male/5 female) suffering from FSHD were scanned on a clinical 3T MRI. The protocol included dynamic MRI5 of the quadriceps on both legs. Clinical Severity Scores (CSS)4 and dynamometry measurements (force peak measured in kilos) were acquired, as well as 6-minutes walking test (6MWT).
For the MRI acquisition, a commercial electrical muscle stimulation (EMS) device was synchronized with a three-directional high-temporal-resolution cine phase contrast (PC) velocity encoding acquisition at 3T5. The stimulator electrodes were placed prior to the scan by identifying the motor point, and the current was set to a sufficient level to evoke muscle twitching without knee extension. Occasionally, this level was reduced at the moment of the MR acquisition because of increased patient discomfort. During periodic contraction of the quadriceps muscle group, a parasagittal slice was acquired with voxels of 2.3x2.3x5 mm3 and a temporal resolution of 42 ms. The velocity encoding had a VENC of 25 cm/s (TR/TE=10.6/7.21 ms, bandwidth/ pixel = 400 Hz/Px, flip angle=10°, FOV=225x300 mm2, 1 k-space line per segment, acquisition time 5 min) and 94 temporal phases were acquired. Each contraction cycle lasted 5 s (1 s ramp-up, 1 s contraction, 1 s ramp-down, 2 s relaxation). Strain rate and strain vectors (over the whole vastus lateralis and vastus intermedius) were extracted from the velocity fields5,6 and normalized by the value of stimulation current for comparison. Texture analysis (gray level co-occurrence matrices (i.e., contrast properties)) was performed on the strain and strain rate maps. Finally, the rate of decay of the strain after removal of the stimulus was estimated from the time curves by fitting of a sigmoid curve.
Pearson’s correlation coefficients were calculated between the MR quantities and the clinical indices (dynamometry and 6MWT).
Both dynamometry variables and 6MWT were acquired for 10 patients (3 extra only had dynamometry). Dynamic MR data were successfully acquired in all cases.
Our data did not show significant correlation of the CSS classification with the quantitative indices of the dynamic acquisitions. However, the patients with lower CSS scores (i.e. less affected by the disease) generally exhibited a larger activated area (Figure 1a). From the analysis of 2D velocities, instead of two distinctive peaks around the moment of the beginning of the stimulation and the moment of release, some intermediate activation was also occasionally present or even a 3rd distinctive peak (Figure 1b). Occasionally though for higher CSS, there was absence of distinctive peaks.
In the strain maps (Figure 2), at the point of highest strain, maximum “activation” was either along the muscle length or localized and in many cases the two legs exhibited very different patterns. Similar effects were seen on the strain rate maps (see Figure 3). While the first strain rate peak map (see Figure 3) provided similar information as the maximum strain, the second peak (at the time point of muscle released from the contraction) yielded different localized information.
Regarding the physical examination, weak correlations of strain and strain rate values with 6MWT (see Figure 4) were found, while correlations were stronger between the 6MWT results and the contrast indices. Finally, the decay rates correlated with the dynamometry with r = 0.61 (Figure 5).