For RF and BF, SENIAM recommendations were used (Hermens et al., 1999). Data was recorded at a sample rate of 2000 samples/s with a multichannel Porti5 EMG system (TMS-international, Enschede, The Netherlands; Hu et al., 2010a). Four clusters of three LED Markers each were fixed onto small lightweight custom-made triangular frames, and attached halfway along the upper and lower legs for registration with a 2 × 3 camera system (OPTOTRAK 3020, Northern Digital, Waterloo, Ontario, Canada), connected via a synchronization cable to the Porti5 EMG system. To

determine leg movements, the heights of the centers of the clusters were calculated. The kinematic sampling frequency was 50 samples/s. The ASLR was performed in supine position with the

feet 20 cm apart (Mens et al., 2001). Subjects were instructed to raise one leg until the heel was 20 cm above the table, without bending the knees, and keeping the leg elevated PD0325901 cell line for about Epacadostat 10 s (“Normal”). To increase statistical precision, this was done three times per leg per condition. After every ASLR, subjects were asked to relax for approximately 10 s. The whole procedure was repeated with a weight added just above the ankle (“Weight”), so that the static moment of the leg with respect to the hip was increased by 50%. To calculate the required amount of weight (Zatsiorsky, 2002; p. 605), manually measured lower extremity anthropometry was used. Finally, the ASLR was repeated with a non-elastic pelvic belt (“Belt”; 3221/3300, Rafys, Hengelo, The Netherlands),

just below the ASIS (Damen et al., 2002; Mens et al., 2006), with a tension of 50 N (Vleeming et al., 1992; Mens et al., 1999), fine-tuned with an inbuilt gauge. Data was analyzed with MATLAB 7.4 (The Mathworks, Natick, MA, USA). Kinematic data were filtered with a 4th order bi-directional low pass Butterworth filter with a cutoff frequency of 5 Hz. We determined the onset and the peak of leg raise, i.e., the first point with zero velocity before/after a peak in velocity. Leg raise velocity was calculated as the height of peak position divided by the time to reach peak position. Due to technical problems with the amplifier, TA EMG was not usable in four subjects, DOK2 which left twelve valid datasets for TA. EMG data were high-pass filtered at 250 Hz (1st order Butterworth; Hu et al., 2010a), then full-wave rectified, and low-pass filtered at 5 Hz (2nd order Butterworth). The median amplitude during ASLR plateau (5 through 10 s after movement onset) was calculated. To quantify the asymmetry of activity of TA, OI, and OE, an Asymmetry Index was calculated as: (ipsilateral − contralateral) activity/(ipsilateral + contralateral) activity × 100%, “ipsilateral” and “contralateral” referring to the leg being raised. Positive values indicate more ipsilateral, negative values more contralateral muscle activity. Outliers were identified from box plots (Figs.