Brain-Computer Interface Controlled Robotic Gait Orthosis

An H. Do, Po T. Wang, Christine E. King, Sophia N. Chun, Zoran Nenadic

Seminal August 2012

TL;DR An EEG brain-computer interface switches a treadmill robotic gait orthosis between standing and walking when the user imagines walking. One able-bodied subject and one with paraplegia from spinal cord injury both controlled it on the first try, with 0.812 average correlation to the cues and no missed walk commands.

What they do

People with paraplegia from spinal cord injury (SCI) cannot walk, and the authors’ earlier EEG work only moved an avatar in virtual reality. Here a noninvasive EEG brain-computer interface (BCI) drives a commercial robotic gait orthosis (a treadmill-mounted Lokomat with about 85% body-weight unloading). Imagining walking starts the orthosis walking; relaxing stops it.

How it works

Training data. 64-channel EEG is recorded while the subject, standing in the orthosis, alternates 30-s epochs of walking imagery and idling for 10 min.
Decoder. Power spectra in 2-Hz bins are reduced to one feature with classwise PCA plus approximate information discriminant analysis, then classified by a linear Bayesian classifier. Online, 0.75-s windows are decoded every 0.25 s.
State machine. The walking probability is averaged over 2 s and compared with two thresholds set from a ~5-min calibration histogram. Arduino-based mouse emulation sends the commands to the orthosis’s user interface.
Checks. EMG and a leg gyroscope test whether the able-bodied subject moved voluntarily. A Monte Carlo autoregressive model of random posteriors gives a chance level.

Results

Offline accuracy was 94.8% (Subject 1) and 77.8% (Subject 2, T6 SCI), against 50% chance.
Over five 5-min online sessions each, the correlation between cues and orthosis walking averaged 0.812 ± 0.048, against 0.438 and 0.498 for simulated chance (p < 10^-4). There were no omissions and 0.8 false alarms per session.
Subject 1’s EMG during BCI walking differed from active and cooperative walking (p < 10^-13) and did not differ from passive walking (p = 0.37).
Peak correlation came at an average lag of 7.4 s, mostly from the orthosis’s locked-in ~5-s power-down sequence.

Limitations

Only two subjects, a treadmill-suspended orthosis and a binary walk/idle command, with no turning or speed control. Subject 2 still had false alarms in the last session, which the authors note could cause harm overground. Robustness to talking or head turning while walking was not formally tested.

Videos

Figures

From Figure 1: A subject suspended in the robotic gait orthosis, wearing an EEG cap, surface EMG electrodes and a gyr... — From Figure 1: A subject suspended in the robotic gait orthosis, wearing an EEG cap, surface EMG electrodes and a gyroscope on the left leg, with the BCI computer driving the orthosis.

From Figure 2: Feature extraction maps for both subjects, one per class; scalp areas near +1 or -1 best separate idli... — From Figure 2: Feature extraction maps for both subjects, one per class; scalp areas near +1 or -1 best separate idling from walking imagery, in the 8-10 Hz bin for Subject 1 and the 10-12 Hz bin for Subject 2.

From Figure 4: A representative online session per subject: instructional cues (blue), decoded BCI state (red, thick ... — From Figure 4: A representative online session per subject: instructional cues (blue), decoded BCI state (red, thick = walking), leg movement from the gyroscope and, for Subject 1, leg EMG.

Abstract

Reliance on wheelchairs after spinal cord injury (SCI) leads to many medical co-morbidities. Treatment of these conditions contributes to the majority of SCI health care costs. Restoring able-body-like ambulation after SCI may reduce the incidence of these conditions, and increase independence and quality of life. However, no biomedical solution exists that can reverse this lost neurological function, and hence novel methods are needed. Brain-computer interface (BCI) controlled lower extremity prosthesis may constitute one such novel approach. One subject with able-body and one with paraplegia due to SCI underwent electroencephalogram (EEG) recording while engaged in alternating epochs of idling and walking kinesthetic motor imagery (KMI). These data were analyzed to generate an EEG prediction model for online BCI operation. A commercial robotic gait orthosis (RoGO) system (treadmill suspended), was interfaced with the BCI computer. In an online test, the subjects were tasked to ambulate using the BCI-RoGO system when prompted by computerized cues. The performance of this system was assessed with cross-correlation analysis, and omission and false alarm rates. The offline accuracy of the EEG prediction model averaged 86.3%. The cross-correlation between instructional cues and BCI-RoGO walking epochs averaged 0.812 +/- 0.048 (p-value<10^-4). There were on average 0.8 false alarms per session and no omissions. This is the first time a person with parapegia due to SCI regained basic brain-controlled ambulation, thereby indicating that restoring brain-controlled ambulation is feasible. Future work will test this system in a population of individuals with SCI. If successful, this may justify future development of invasive BCI-controlled lower extremity prostheses. This system may also be applied to incomplete SCI to improve neurological outcomes beyond those of standard physiotherapy.

Summaries are written automatically from each paper's text. Figures and videos come from the paper and its project page; all credit goes to the authors.

Matt Strong