Treadmill-based system for postural studies: Design and validation

Abstract

Computer-controlled treadmills are common in many gait labs and offer great potential for conducting perturbation-based postural studies. However, the time-course of these disturbances can be too brief to be controlled manually through product software. Here we present a system that combines a Bertec® split-belt treadmill with custom hardware and software to deliver postural disturbances during standing and record data from multiple sources simultaneously. We used this system to administer to 15 healthy participants an 8-session perturbation-based training protocol in which they learned to respond without stepping to progressively larger perturbations. Kinematic, electromyographic, and force data were collected throughout. Motion capture was used to characterize the accuracy and repeatability of the treadmill-delivered perturbations with respect to duration, displacement, and peak velocity. These (observed) data were compared to that expected based on software commands and the known constraints of the treadmill (i.e., 10 Hz operating speed). We found perturbation durations to be as expected. Peak velocities and displacements were slightly higher than expected (average increases were 0.59 cm/s and 1.76 cm, respectively). Because this increase in magnitude was consistent, it did not impede training or affect data analysis. Treadmill behavior was repeatable across 95 % of trials.

Keywords: Computer-controlled treadmill; Fall prevention; Perturbation-based training; Postural control; Rehabilitation.

Figure 1:

System Configuration and Hardware. A.) Commercial hardware includes a split-belt treadmill, wired EMG system, display monitor and motion capture technology. B.) A low-cost custom-built tachometer, constructed from the electronics of a disassembled optical mouse and an Arduino Uno, is attached to the treadmill directly below the belt. Movement detection is based on two software thresholds: (1) >5 units absolute displacement; and (2) >3 consecutive events in same direction; it is relayed to custom software via USB. C.) Schematic diagram: The investigator controls equipment and data collection through custom software that interacts with standard product software. Solid red lines indicate command pathways; dotted blue lines indicate data pathways. Perturbation commands, programmed in custom software, are sent to the treadmill via product software and hardware and regulated through timers triggered by tachometer data. Pre-programmed instructions are displayed on a monitor located in front of the participant. Motion capture technology, triggered via an external data acquisition (DAQ) device, is used to collect kinematic, electromyographic (EMG), and force data. A treadmill emergency stop button that bypasses all software is located next to the investigator.

Figure 2:

Custom Software. A.) The parameters defining perturbation type and difficulty levels are programmable. Currently all perturbations follow a trapezoidal-shaped velocity profile with difficulty levels assigned based on peak velocity. Acceleration and deceleration time periods are 200 ms each, with rates equal in absolute value; time at peak velocity varies from 200–300 ms. Thus, perturbation duration varies from 600–700 ms. B.) The graphical user interface (GUI) enables the investigator to select the perturbation type and level, to initiate perturbation trials, and to record the subject, session, and perturbation type and level administered for each trial. Experimental run files (i.e., pre-defined sets of trials) can be imported from comma-separated values (CSV) files; changes to the imported file can be made on a trial-by-trial basis if desired. Exported files are used to align perturbation information with data recorded through the motion capture system. C.) This flowchart shows the actions triggered by software commands once the start perturbation button is activated through the GUI. Once the acceleration command is sent, treadmill ramp-up and ramp-down durations are controlled through 2 additional mechanisms: the timer triggered by tachometer data, and input to the commercial software regarding velocity. During the acceleration phase (ramp-up) a 200 ms timer is triggered by the initial movement of the treadmill belts, thereby eliminating the problem of not knowing exactly when in the 100 ms window (the treadmill controller operates at 10 Hz) the acceleration command will be executed. During the deceleration phase (ramp-down) the treadmill is commanded to decelerate to zero (stop) at a rate which takes 200 ms to complete based on its peak velocity (these data are stored in the internal database of the custom software, for each difficulty level). For safety, an additional timer, activated when the acceleration command is sent, monitors the time between trial initiation and movement of the treadmill belt as detected by the tachometer. If belt movement is not detected within the allotted timeframe, the treadmill stops and an error message is generated. Participant instructions are appropriately timed to treadmill movement. Instruction screens appear sequentially at time points indicated in the flowchart by the numbers in parentheses. The green circle informs the participant that the treadmill belts will move within the next 2–4 s; s/he is pre-instructed to focus on the crosshairs. The red circle indicates that the perturbation trial is over.

Figure 3:

Perturbation-based Training Protocol. A.) Our current training protocol consists of one baseline assessment session (BL), six training sessions (T#1-T#6), and one final assessment session (F). Stepping threshold, defined as the level where the size of the perturbation is large enough to evoke stepping on three consecutive trials, is determined at the beginning of each session. In T#1-T#6, stepping threshold determination is followed by 60 trials of practice, with levels increasing as performance improves. B.) This figure depicts a typical training session. Perturbation levels range from 1–20 with higher values indicating faster larger perturbations. Trials 1–13 depict stepping threshold determination; here the participant begins at perturbation level 1 (the lowest level of difficulty) and moves up one level at a time until s/he steps during three consecutive trials. Trials 14–73 depict progressive practice; here the participant begins one level below stepping threshold, progresses to the next higher level after completing three successful (non-stepping) trials, and moves down one level if s/he steps. C.) Parameters associated with perturbation difficulty level. Displacement is calculated based on the programmed acceleration/deceleration rate and time, and the expected rate and time at peak velocity. Minimum and maximum displacements correspond to 200 ms and 300 ms at peak velocity, respectively. The expected displacement range is equal to its median value ±11%.

Figure 4:

Measurement of Perturbation Duration and Size. Treadmill belt movement recorded through motion capture technology was used to measure the duration, displacement, and peak velocity of the actual perturbations. Data observed through motion capture (shown in green) are displayed alongside that expected (shown in black) based on software commands. Figures A and B show a representative sample collected during a Level 10 perturbation. A.) For each trial, perturbation duration (shown as time in ms) was calculated by subtracting time at perturbation onset from that at termination; duration was expected to be between 600–700 ms. Perturbation displacement was computed by subtracting marker position at perturbation onset from that at termination. For a L10 perturbation, displacement was expected to range from 20–25 cm; the average displacement observed at this level was 24.7 cm. B.) Peak velocity (Vpeak) was computed by averaging the velocity observed across the middle 96 ms of the trial. For a perturbation at L10, peak velocity was expected to be 50.00 cm/s; as shown here, the average peak velocity observed at this level was 50.7 cm/s. C.) The perturbation size expected for each level (L) is indicated by the black line; the length of this line depicts the range of displacements possible at each level based on the expected variation in perturbation duration. Each green circle represents the actual size (i.e., peak velocity and displacement) of a single trial as observed through motion capture. As shown here, the actual perturbation size (as observed through motion capture) was slightly larger than that expected based on software commands. Note: Stepping threshold varied across participants, therefore higher perturbation levels contain fewer trials.

Figure 5:

Accuracy and Consistency of Treadmill Perturbations. To determine the accuracy and repeatability of treadmill perturbations, data observed through motion capture were compared to that expected based on software commands. For Figures 5A, 5B, 5C, 5E, and 5F, each blue circle represents data from a single trial; observed values were computed from motion capture data; expected values were computed based on software commands. Note: Stepping threshold varied across participants, therefore higher perturbation levels contain fewer trials. For Figure 5D each circle represents the average of 3 observed trials. A.) Perturbation duration was expected to range between 600 and 700 ms; as shown by the shaded area, 97% of trials occurred within this timeframe. Mean perturbation duration observed across all trials was 637 ms. B.) The average difference between observed and mean expected displacements was 1.76 cm, for grouped data. As expected, displacement range increased with level. C.) Perturbation displacements were expected to vary in accord with duration variability. As shown by the shaded area, 95% of observations were within the expected range of mean ±11%. D.) Peak velocity was averaged for 3 trials/level for each of the 6 participants who reached perturbation Level 15; no effect of participant mass on peak velocity was observed. E.) The group average difference between observed and expected peak velocities was 0.59 cm/s. F.) For 95% of trials, peak velocities were within ±0.8 cm/s of the mean (shaded area).

Figure 6:

Data Sample. A.) Stepping threshold data from a single participant is shown, by session. For this participant, stepping threshold (i.e., the perturbation difficulty level that causes a participant to step on three consecutive trials) increased progressively across six training sessions. B.) The motion capture software was used to record data collected from three different sources. Kinematic data collected from markers placed directly on the treadmill belt synchronize participant electromyographic and ground-reaction-force (GRF) data to the perturbation.

Figure 7:

Effect of Perturbation-based Training on Stepping Threshold. Fifteen participants (P) completed the perturbation-based training protocol. For 14 participants, stepping threshold, defined as the perturbation level that evoked a stepping response on three consecutive trials, increased from baseline to final. *Three participants (P09, P10, and P13) performed identically (Baseline =10, Final =13). P14 was the only participant whose stepping threshold did not change.