Towards effective non-invasive brain-computer interfaces dedicated to gait rehabilitation systems

Abstract

In the last few years, significant progress has been made in the field of walk rehabilitation. Motor cortex signals in bipedal monkeys have been interpreted to predict walk kinematics. Epidural electrical stimulation in rats and in one young paraplegic has been realized to partially restore motor control after spinal cord injury. However, these experimental trials are far from being applicable to all patients suffering from motor impairments. Therefore, it is thought that more simple rehabilitation systems are desirable in the meanwhile. The goal of this review is to describe and summarize the progress made in the development of non-invasive brain-computer interfaces dedicated to motor rehabilitation systems. In the first part, the main principles of human locomotion control are presented. The paper then focuses on the mechanisms of supra-spinal centers active during gait, including results from electroencephalography, functional brain imaging technologies [near-infrared spectroscopy (NIRS), functional magnetic resonance imaging (fMRI), positron-emission tomography (PET), single-photon emission-computed tomography (SPECT)] and invasive studies. The first brain-computer interface (BCI) applications to gait rehabilitation are then presented, with a discussion about the different strategies developed in the field. The challenges to raise for future systems are identified and discussed. Finally, we present some proposals to address these challenges, in order to contribute to the improvement of BCI for gait rehabilitation.

Figure 1.

Illustration of different phases of the gait cycle. (A) New gait terms; (B) classic gait terms; and (C) Percentage of gait cycle. Note: this figure is adapted with permission from [16]; Copyright Demos Medical Publishing Inc., 2004.

Figure 2

Influence of walking speed on joint trajectories. Joint trajectories of the (A) hip, (B) knee and (C) ankle joint at 10 different walking speeds (this figure is reproduced with permission from [17]; Copyright IOS Press, 2010).

Figure 3

Influence of walking speed on electromyographic (EMG) activity patterns. EMG activity patterns during different walking speeds of the (A) gluteus maximus (GL); (B) rectus femoris (RF); (C) vastus lateralis (VL); (D) vastus medialis (VM); (E) lateral hamstrings (HL); (F) medial hamstrings (HM); (G) tibialis anterior (TA); and (H) gastrocnemius medialis (GM) muscles. EMG signals were normalized for each subject and each condition by setting the difference between the lowest and highest EMG amplitude at 100% and normalizing the curve according to this value (this figure is reproduced with permission from [17]; Copyright IOS Press, 2010).

Figure 4

Model of the different pathways indicating how afferents can act on the central pattern generator (CPG) during the stance phase of locomotion. The CPG contains a mutually inhibiting extensor and flexor half-center (EHC and FHC, respectively). During the stance phase, the load of the lower limb is detected by group I extensor muscle afferents and group II (low threshold) cutaneous afferents, which activate the EHC. In this way, extensor activity is reinforced during the loading period of the stance phase. At the end of the stance phase, group Ia afferents of flexor muscles excite the FHC (which inhibits the EHC) and, thereby, initiate the onset of the swing phase (this figure is reproduced with permission from [42]; Copyright Elsevier, 1998).

Figure 5

Global view of the human locomotion machinery (this figure is reproduced with permission from [63]; Copyright Springer-Verlag, 2005; see text for details).

Figure 6

Comparison of real (PET) and imagined locomotion (fMRI) brain activations (this figure is reproduced with permission from [105]; Copyright Elsevier, 2010). Sagittal midline and render views are shown. It can be seen that during real locomotion, the primary motor sensory cortices (pre- and post-central gyri) are active (left) as compared to the supplementary motor areas (superior and medial frontal gyri) in mental imagery of locomotion (right). Furthermore during imagined locomotion, the basal ganglia (caudate nucleus, putamen) are active, which is not the case for real locomotion.

Figure 7

Illustration of the executive and planning networks of locomotion, as suggested in [105]. Execution of locomotion in a non-modulatory steady state (left side) goes from the primary motor cortex areas directly to the spinal central pattern generators (CPG), thereby bypassing the basal ganglia and the brainstem locomotor centers. A feedback loop runs from the spinal cord to the cerebellum and, thereby, via the thalamus to the cortex. For planning and modulation of locomotion (right side), cortical locomotor signals originate in the prefrontal supplementary motor areas and are transmitted through the basal ganglia via disinhibition of the subthalamic locomotor region (SLR) and mesencephalic locomotor region (MLR), where they converge with cerebellar signals from the cerebellar locomotor region (CLR). The MLR functionally represents a cross point for motor information form basal ganglia and cerebellar loops. Descending anatomical projections are directed to the medullary and pontine reticular formations (PMRF) and the spinal cord; ascending projections are in the main part concentrated on the basal ganglia and the non-specific nuclei of the thalamus (not shown for the sake of clarity). The CLR also projects via the thalamus back to the cortex. Cortical signals are furthermore modulated via a thalamo-cortical-basal ganglia circuit. The schematic drawing shows a hypothetical concept of a direct pathway of steady-state locomotion (left) and an indirect pathway of modulatory locomotion (right). SMA, supplementary motor cortex. This figure is reproduced with permission from [105]; Copyright Elsevier, 2010.

Figure 8

Time-frequency plots (wavelet transformation) of LFP oscillations during gait cycle. Upper row (A) and (B): analyzed electrode pair. The right electrode pair is on the right side. (C) and (D): goniometer traces. Modulation of LFPs occurs in the 6–11 Hz frequency range. In this frequency band, amplitudes are upregulated during the early stance phase and swing phase of the contralateral leg. LL: left leg; RL: right leg; Gonio: goniometer; Flex: flexion. This figure is reproduced with permission from [142]; Copyright Elsevier, 2011.

Figure 9

General scheme of a classical brain-computer interface (BCI): first of all, the subject performs a specific mental task in order to produce a signal of interest in his brain; then, this signal is acquired and generally pre-processed in order to get rid of different artifacts. Afterwards, some discriminating features are extracted and classified (pattern recognition) to determine which specific signal was produced. Finally, the identified signal is associated with a specific action to be performed by a computer or any electronic device.

Figure 10

When looking at a Bereitschaftspotential (BP) signal, three main sections are observed: no potential, a slow decreasing potential early BP and a steeper late BP (this figure is reproduced with permission from [154]; Copyright Elsevier, 2005). MRCP is the movement-related cortical potential.

Figure 11

It clearly appears that the BP potentials are similar for all five tasks. A classification between those tasks would be difficult. The potentials are strong over the motor cortex area close to the midline (MS is the movement start). This figure is reproduced with permission from [108]; Copyright Elsevier, 2005.