2.6. Signal transmission, data processing and normalization

Since all force sensor signals from treadmill, stairway and rope-pulley system were recorded together with the EMG signals in one LabVIEW-based software (IMAGO Record, pfitec, Endingen, D), no delay due to synchronization was present. Therefore, no later synchronization during the post-processing was needed.

First, EMG signals were transmitted across a differential-preamplifier to a telemetric main amplifier (PowerPack, pfitec, Endingen, D), band-pass filtered at 10 Hz–1000 Hz and recorded at 2000 Hz for treadmill walking and stair descent, and at 4000 Hz for the stretch-reflex measurement [20]. Afterwards, the signals were converted from analogue to digital (type NI PCI 6255, National Instruments, Austin, USA), registered and further processed with the same LabVIEW-based software (IMAGO Record, pfitec, Endingen, D). Then, all raw EMG signals were full wave rectified. Additionally, raw EMG signals from treadmill walking and stair descent were band-pass filtered at 10–500Hz (Butterworth, 2nd order). For the defined time windows as described above, RMS values were calculated and exported in Excel spreadsheets (Windows 10, Microsoft Corporation, Redmond WA, USA). Individual means out of 30 gait cycles (treadmill walking), ten steps (stair descent), and 30 tibia translations per extremity for each muscle in each time interval were calculated. RMS values were normalized according to the corresponding time intervals retrieved during treadmill walking (100 % of neuromuscular activity) and used for comparability within subjects. Reflex responses during artificial tibial translation, and activations during stair descent were normalized and expressed as a percentage (% EMG) of respective treadmill walking activity [21,22].