Oqus7

The IMUs were synchronized with a twelve-camera optoelectronic system (Oqus7+, Qualisys Göteborg, Sweden) according to the protocol described in a previous paper whose goal was to compare gait parameter outputs from the clinical silver standard and the wearable system (Carcreff et al., 2018 (link)). In the present study, the optoelectronic system was only used to automatically crop continuous IMUs’ data into several straight gait trials (to discard turns and standing periods).

A 12-camera motion capture system (Oqus 7+, Qualisys, Göteborg, Sweden), set at a 100 Hz sampling frequency, was used to track cutaneous reflective markers. The participants were equipped with 35 anatomical markers (14 mm diameter) placed on the whole body, according to the Conventional Gait Model 1.0^{39 (link),40 (link)} (Fig. 1).Figure 1

Summary of markerset and calculated variables. Red and yellow markers: Reflective markers placed according to the Conventional Gait Model 1.0.

The gait analysis measurement started with a 10-s recording of the participant standing upright (T-pose). Then, participants were asked to walk barefoot back and forth on a 10-m walkway at three different self-selected speeds: slow, comfortable, and fast. Walking trials at each speed were repeated three times.

Participants were instructed to walk barefoot at a comfortable self-selected speed along a 10 m walkway. Kinematic parameters were measured using a 12-camera motion analysis system (model Oqus 7+, Qualisys, Göteborg, Sweden) between 2015 and 2019, a 12-camera motion analysis system (Vicon MX3+, Vicon Peak, Oxford, United Kingdom) between 2007 and 2015 and a 6-camera motion analysis system (Vicon 460, Vicon Peak, Oxford, United Kingdom) before 2007. The marker trajectories were recorded at 100 Hz and filtered using the predicted mean-squared error filter MSE10 in the Nexus software before 2015 and high-pass 4th order Butterworth filter (10 Hz) after. Participants were equipped with 35 reflective markers placed on the skin at defined anatomical and technical landmarks according to the full-body Plug-in-Gait model (Davis et al., 1991 ).

The workflow of the study is represented in Fig 1. The anthropometric data (leg length, knee and ankle width, height and weight) were collected by experienced physiotherapists. Then, participants were equipped according to the CGM marker set [12 ] (12.5mm) and asked to walk barefoot at a self-selected speed. A 12-camera motion capture system (Oqus7+, Qualisys, Göteborg, Sweden) tracked the marker trajectories at 100Hz. Gait kinematic was processed with the Vicon PiG clone, provided as CGM1.1 by the open-source library PyCGM2, which requires a static trial for calibration [21 (link)]. In agreement with the original version of the CGM [14 ], the coronal plane of the femur was constructed from the hip joint centre, the KNE and the lateral thigh wand mounted marker. A systematic offset was introduced to the KNE marker along the AP and PD axis of the thigh to create a virtual marker. Virtual markers were placed every 45° around the original position at five different magnitudes (distance from the original marker): 5, 10, 15, 20 and 30 mm leading to a total of 40 virtual misplacements of the KNE position for each patient. The Eq (1) estimates the new position of the KNE marker (KNE_misp) as a result of the sum of the original position (KNE_ori) on the segment coordinate system with an error (E) defined in function of magnitude (ɛ) and direction (θ).