Figure 1 - uploaded by Matthieu Duvinage
Content may be subject to copyright.
![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.](profile/Matthieu-Duvinage/publication/259632493/figure/fig1/AS:297139977637888@1447855099124/llustration-of-different-phases-of-the-gait-cycle-A-New-gait-terms-B-classic.png)
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.
Source publication
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 exp...
Context in source publication
Context 1
... harmful consequences, like recurrent pain and injuries at the interface between their residual limb and the prosthesis. Active prostheses solve these problems partially: powered by a battery-operated motor, they move on their own and, therefore, reduce the fatigue of the amputees, while improving their posture. Two main categories of active prostheses exist to date. Firstly, by analyzing the motion of the healthy leg or the upper-body by means of sensors, the control system can identify the phase of the gait cycle and trigger an actuator to appropriately adjust one or more prosthetic or orthotic joint [8–12]. The second type of active prostheses (or orthoses) is controlled by myoelectric signals recorded at the surface of the skin, just above the muscles. These signals are then used to guide the movement of the artificial limb [13–15]. Although the improvement brought by active prosthetic technology with respect to conventional prostheses is indisputable, an intuitive interface from which users intent can be determined is still missing. The purpose of this paper is to review, firstly, the substantial progress made in the understanding of human locomotion control and, in the second part, the exploitation of this knowledge that is being made in order to develop non-invasive brain-computer interfaces dedicated to walk rehabilitation systems. Section 2 summarizes the main mechanisms involved in human locomotion control. Section 3 focuses on the description of supra-spinal control of locomotion by summarizing the knowledge acquired to date thanks to multiple methods of measuring neuronal activity. Section 4 discusses different strategies developed to produce walk rehabilitation systems driven by non-invasive brain-computer interfaces. Accumulating evidence suggests that human locomotion is actually based on a very complex hierarchical system, which includes several control networks located both at the spinal and supra-spinal levels. Basically, high-level motor commands are sent by the brain to a spinal network composed of central pattern generators (CPGs), and at the same time, each level of motor control receives and transmits peripheral sensory information (sensory feedback), which is used to modify the motor output at that level. This section is first devoted to the description of each level of locomotor control, including arguments supporting the existence of a CPG network and, simultaneously, the permanent action of supra-spinal control. Then, the focus is on the spatial organization of supra-spinal control and its temporal characteristics. Human walking is composed of successive periodic and symmetric movements produced by a precise sequence of collective actions, one leg alternating with the other one. The gait cycle is usually defined as starting with the first contact (initial contact, or heel contact in normal gait) of one foot, so that the end of the cycle occurs with the next contact of the same (ipsilateral) foot (see Figure 1). Each cycle begins with a stance phase (when the foot hits the ground) and proceeds through a swing phase, until the cycle ends with the limb’s next initial contact. The stance phase of gait is divided into four periods: loading response, mid-stance, terminal stance and preswing. The swing phase is divided into three periods: initial swing, mid-swing and terminal swing. The beginning and ending of each period are defined by specific events, listed in Table 1. Figure 2
...
Similar publications
Brain-Computer Interface (BCI) techniques hold a great promise for neuroprosthetic applications. A desirable BCI system should be portable, minimally invasive, and feature high classification accuracy and efficiency. As two commonly used non-invasive brain imaging modalities, Electroencephalography (EEG) and functional near-infrared spectroscopy (f...
Physiological processes—such as, the brain's resting-state electrical activity or hemodynamic fluctuations—exhibit scale-free temporal structuring. However, impacts common in biological systems such as, noise, multiple signal generators, or filtering by transport function, result in multimodal scaling that cannot be reliably assessed by standard an...
Citations
... Depending on the location of electrodes, BCI involves invasive, 20 semi-invasive, 21 and non-invasive BCIs. [22][23][24][25] In addition, EEG-based BCI also contains evoked and spontaneous. The evoked BCI requires external stimuli (i.e., visual, auditory, and sensory stimuli) to elicit a brain response. ...
Stroke has been the second leading cause of death and disability worldwide. With the innovation of therapeutic schedules, its death rate has decreased significantly but still guides chronic movement disorders. Due to the lack of independent activities and minimum exercise standards, the traditional rehabilitation means of occupational therapy and constraint-induced movement therapy pose challenges in stroke patients with severe impairments. Therefore, specific and effective rehabilitation methods seek innovation. To address the overlooked limitation, we design a pneumatic rehabilitation glove system. Specially, we developed a pneumatic glove, which utilizes ElectroEncephaloGram (EEG) acquisition to gain the EEG signals. A proposed EEGTran model is inserted into the system to distinguish the specific motor imagination behavior, thus, the glove can perform specific activities according to the patient's imagination, facilitating the patients with severe movement disorders and promoting the rehabilitation technology. The experimental results show that the proposed EEGTrans reached an accuracy of 87.3% and outperformed that of competitors. It demonstrates that our pneumatic rehabilitation glove system contributes to the rehabilitation training of stroke patients.
... Moreover, we have shown in Figure 5 the MTG activation estimates obtained from the musculoskeletal models for HFE, KFE, and ADPF with experimental surface electromyography (sEMG) data from 12 leg muscles during walking motion. The MTG activation estimates were shown against the experimental data from experimental studies by Castermans et al [69], Wu et al [70], and Ivanenko et al [71]. Note that the MTG activation is generated under the assumption of zero co-contraction, whereas actual sEMG signals reveal some degree of co-contraction resulting from the simultaneous activation of multiple human muscles. ...
Human motion capture technology is utilized in many industries, including entertainment, sports, medicine, augmented reality, virtual reality, and robotics. However, motion capture data only allows the user to analyze human movement at a kinematic level. In order to study the corresponding dynamics and muscle properties, additional sensors such as force plates and electromyography sensors are needed to collect the relevant data. Collecting, processing, and synchronizing data from multiple sources could be laborious and time-consuming. This study proposes a method to generate the dynamics and muscle properties of existing motion capture datasets. To do so, our method reconstructs motions via kinematics, dynamics, and muscle modeling with a musculoskeletal model consisting of 14 joints, 40 degrees of freedom, and 15 segments. Compared to current physics simulators, our method also infers muscle properties to ensure our human model is realistic. We have met International Society of Biomechanics standards for all terminologies and representations. Furthermore, our integrated musculoskeletal model allows the user to preselect various anthropometric features of the human performing the motion, such as height, mass, level of athleticism, handedness, and skin temperature, which are often infeasible to estimate from monocular videos without appropriate annotations. We apply our method on the Human3.6M dataset and show that our reconstructed motion is kinematically similar to the ground truth markers while being dynamically plausible when compared to experimental data found in literature. The generated data (Human3.6M+) is available for download.
... It is also known that human gait consists of continuous periodic and symmetrical movements produced by a precise series of coordinated movements, alternating between one leg and the other [31]. In the human and the proposed method, the contribution ratio of each principal component does not differ significantly for the left and right legs. ...
There are many studies analyzing human motion. However, we do not yet fully understand the mechanisms of our own bodies. We believe that mimicking human motion and function using a robot will help us to deepen our understanding of humans. Therefore, we focus on the characteristics of the human gait, and the goal is to realize a human-like bipedal gait that lands on its heels and takes off from its toes. In this study, we focus on kinematic synergy (planar covariation) in the lower limbs as a characteristic gait seen in humans. Planar covariation is that elevation angles at the thigh, shank, and foot in the sagittal plane are plotted on one plane when the angular data are plotted on the three axes. We propose this feature as a reward for reinforcement learning. By introducing this reward, the bipedal robot achieved a human-like bipedal gait in which the robot lands on its heels and takes off from its toes. We also compared the learning results with those obtained when this feature was not used. The results suggest that planar covariation is one factor that characterizes a human-like gait.
... Gait is a semiautomatic function, which requires attention for locomotor control [1,2]. Attentional demand is lower during normal walking, and higher for difficult ones such as dual-task walking [1,3]. Studies using functional near-infrared spectroscopy (fNIRS), which enables real-time monitoring of brain activation during dual-task walking to confirm the association between prefrontal cortex (PFC) activity and dual-task walking, have reported that dual-task walking leads to greater PFC activation than that with single-task walking [2,4]. ...
Background
Studies using functional near-infrared spectroscopy (fNIRS) have shown that dual-task walking leads to greater prefrontal cortex (PFC) activation compared to the single-task walking task. However, evidence on age-related changes in PFC activity patterns is inconsistent. Therefore, this study aimed to explore the changes in the activation patterns of PFC subregions in different activation phases (early and late phases) during both single-task and dual-task walking in both older and younger adults.
Methods
Overall, 20 older and 15 younger adults performed a walking task with and without a cognitive task. The activity of the PFC subregions in different phases (early and late phases) and task performance (gait and cognitive task) were evaluated using fNIRS and a gait analyzer.
Results
The gait (slower speed and lower cadence) and cognitive performance (lower total response, correct response and accuracy rate, and higher error rate) of older adults was poorer during the dual task than that of younger adults. Right dorsolateral PFC activity in the early period in older adults was higher than that in younger adults, which declined precipitously during the late period. Conversely, the activity level of the right orbitofrontal cortex in the dual-task for older adults was lower than for younger adults.
Conclusions
These altered PFC subregion-specific activation patterns in older adults would indicate a decline in dual-task performance with aging.
... EEG could provide a non-invasive way to BCI with the characteristics of simple structure, high safety, and good realtime performance (Castermans et al., 2013). With the features extracted from the EEG signals, the classification based on the machine learning algorithms can convert them into the control commands for assistive or rehabilitation devices (Li et al., 2016). ...
Background
The aging of the world population poses a major health challenge, and brain–computer interface (BCI) technology has the potential to provide assistance and rehabilitation for the elderly.
Objectives
This study aimed to investigate the electroencephalogram (EEG) characteristics during motor imagery by comparing young and elderly, and study Convolutional Neural Networks (CNNs) classification for the elderly population in terms of fatigue analysis in both frontal and parietal regions.
Methods
A total of 20 healthy individuals participated in the study, including 10 young and 10 older adults. All participants completed the left- and right-hand motor imagery experiment. The energy changes in the motor imagery process were analyzed using time–frequency graphs and quantified event-related desynchronization (ERD) values. The fatigue level of the motor imagery was assessed by two indicators: (θ + α)/β and θ/β, and fatigue-sensitive channels were distinguished from the parietal region of the brain. Then, rhythm entropy was introduced to analyze the complexity of the cognitive activity. The phase-lock values related to the parietal and frontal lobes were calculated, and their temporal synchronization was discussed. Finally, the motor imagery EEG data was classified by CNNs, and the accuracy was discussed based on the analysis results.
Result
For the young and elderly, ERD was observed in C3 and C4 channels, and their fatigue-sensitive channels in the parietal region were slightly different. During the experiment, the rhythm entropy of the frontal lobe showed a decreasing trend with time for most of the young subjects, while there was an increasing trend for most of the older ones. Using the CNN classification method, the elderly achieved around 70% of the average classification accuracy, which is almost the same for the young adults.
Conclusion
Compared with the young adults, the elderly are less affected by the level of cognitive fatigue during motor imagery, but the classification accuracy of motor imagery data in the elderly may be slightly lower than that in young persons. At the same time, the deep learning method also provides a potentially feasible option for the application of motor-imagery BCI (MI-BCI) in the elderly by considering the ERD and fatigue phenomenon together.
... The gamma rays emitted by the tracer are then detected using a CT-Scan. This allows a spatial resolution of about 1mm and a temporal resolution of about 0.2s (Castermans et al., 2013). PET is non invasive but expose patients to ionizing radiations, it should therefore be used only sporadically. ...
Invasive brain-computer interfaces controlled by paralyzed people could restorenatural speech production by providing real-time speech synthesis from corticalactivity. This thesis aims at decoding existing invasive recordings of speech activityin an offline setting, using real-time compatible methods that could later be used ina natural speech brain-computer interface.A focus was made on decoding speech from cortical activity using linear methods,in particular partial least squares regression, which has been successfully used inmotor brain-computer interfaces before but not for speech decoding yet. Two mainapproaches were compared: 1. direct decoding of F0 and mel cepstral coefficientsof speech, and 2. indirect decoding of speech through an articulatory representation.In order to decode articulatory trajectories from cortical activity, those were firstinferred from the patient’s audio recordings using dynamic time warping. Severalfeedforward and recurrent neural networks were trained on a separate electromag-netic articulography dataset to perform articulatory-to-acoustic synthesis, and wereevaluated using objective and perceptive criterions. The best model was finetunedto predict mel cepstral coefficients of speech from decoded articulatory trajectories.Speech was synthesized from decoded F0 and mel cepstral coefficients using anMLSA filter, for both decoding paradigms.Both direct and indirect decoding of acoustic features of speech achieved signifi-cant speech decoding with similar performances, although not intelligible. Partialleast squares regression was found to perform a more efficient feature reductionthan PCA-based linear regressions, for a similar performance. Prior to decoding,noisy channels and spectral features of cortical activity that do not contain speechinformation were successfully removed using an automatic feature selection. It wasfound that decoding from spectrograms of cortical activity was best when using allselected frequencies up to 200Hz and concatenating the last 200 ms of brain activity.Decoding of articulatory trajectories was significantly better from frontal electrodesthan from temporal electrodes, and the opposite was true for acoustic features ofspeech. However, in both cases decoding was significantly better when includingall electrodes. Finally, our experiments suggest that decoding could be improved bysplitting a speech decoder into a voicing classifier and a regression-based decoderonly active on voiced segments.vIn this thesis, we set up an entire real-time-compatible decoding pipeline based onlinear methods. It should now be implemented for further evaluation in a close-loopexperiment. Meanwhile, although decoding was much better than chance, linearmethods are likely not good enough yet for a brain-computer interface generatingnatural speech. Further work should focus on developing real-time compatibledecoders based on other methods like deep neural networks.
... Human gait is a biometric attribute that is useful and attracting attention in different fields such as surveillance, biomedical engineering, clinical analysis, etc. Commonly, gait analysis is essential in a clinical investigations such as fall detection [1], rehabilitation [2,3], physical therapy [4], etc., for the well-being of a patient suffering from underlying diseases such as strokes, Parkinson's, or progressive supranuclear palsy (PSP). Current studies focus on the recent development of human gait rehabilitation therapy based on the state of the brain by employing the brain-computer interface (BCI) system [5][6][7]. ...
Smartphone-based gait recognition has been considered a unique and promising technique for biometric-based identification. It is integrated with multiple sensors to collect inertial data while a person walks. However, captured data may be affected by several covariate factors due to variations of gait sequences such as holding loads, wearing types, shoe types, etc. Recent gait recognition approaches either work on global or local features, causing failure to handle these covariate-based features. To address these issues, a novel weighted multi-scale CNN (WMsCNN) architecture is designed to extract local to global features for boosting recognition accuracy. Specifically, a weight update sub-network (Ws) is proposed to increase or reduce the weights of features concerning their contribution to the final classification task. Thus, the sensitivity of these features toward the covariate factors decreases using the weight updated technique. Later, these features are fed to a fusion module used to produce global features for the overall classification. Extensive experiments have been conducted on four different benchmark datasets, and the demonstrated results of the proposed model are superior to other state-of-the-art deep learning approaches.
... Since fNIRS systems are vulnerable to different noise resources, there must be artefacts removal systems like independent component analysis (ICA) or wavelet. Generally, fNIRS is portable and inexpensive technology; therefore, it can use outside the lab environment [Castermans et al., 2014]. ...
Biometrics are getting more popular in the field of security systems and authentication. This is because biometrics are less able to be lost and less able to be stolen or spoofed. EEG-based biometrics are getting more attention recently since they are more resistant to be hacked. This thesis aims to design and implement techniques for the secure authentication of users based on electroencephalogram (EEG) signals. The study was conducted using three datasets: Simultaneous Task EEG workload Dataset, EEG Alpha wave Dataset, and Local Dataset.
The first problem addressed is the curse of dimensionality. Four reduced feature sets were used to reduce the dimensions of the systems, namely, cluster map, ANOVA F-Value, logistic regression weights; the cluster map method reached the highest performance with an 82.37\% reduction in computation time.
The second problem is to reduce the time required to record EEG signals. Different scenarios with different EEG recording durations were tested. The results reveal a temporal threshold, equals to 4 seconds, that balances between performance and implementability. \\
The third problem is the effect of the auditory stimuli. To do so, six experiments were conducted, native, non-native, and neutral songs. The three songs were conducted using In-Ear and Bone-Conducting headphones. The results show that an increase in the performance of authentication equals 9.27\% when using auditory stimuli. Additionally, it shows that using In-Ear or Bone-Conducting auditory stimuli is based on the balance between performance and implementability. Finally, the performance of EEG-based authentication is independent of the language of auditory stimuli.
In conclusion, this thesis contributes to the development of EEG-biometrics by bridging some important gaps in the field.
... Thus, the segmented data were time-warped and averaged together for all strides, so that the stationary (S1, 10-30% of gait cycle, and S2, 60-80% of gait cycle) and transition phases of the gait cycle (T1, 30-60% of gait cycle, and T2, 80-10% of gait cycle) occurred at the same times, as well as the HS and toe-off (TO) of the affected and unaffected lower limb (Calabrò et al. 2017a, b;Light et al. 2010). The spectrum analysis was carried on delta (2-4 Hz), theta (4-7 Hz), alpha (8-12 Hz), beta (12-30 Hz), and gamma (31-45 Hz) frequency bands in relation to the sub-phases of the gait cycle (Castermans et al. 2014;Pfurtscheller and Neuper 1994). Thereafter, we calculated the electrode group-wise effective connectivity by computing the re-normalized partial directed coherence as a causal estimator derived from the AMVAR model coefficients (Sharma et al. 2013). ...
Patients with stroke can experience a drastic change in their body representation (BR), beyond the physical and psychological consequences of stroke itself. Noteworthy, the misperception of BR could affect patients' motor performance even more. Our study aimed at evaluating the usefulness of a robot-aided gait training (RAGT) equipped with augmented visuomotor feedback, expected to target BR (RAGT + VR) in improving lower limb sensorimotor function, gait performance (using Fugl-Meyer Assessment scale for lower extremities, FMA-LE), and BR (using the Body Esteem Scale—BES- and the Body Uneasiness Test—BUT), as compared to RAGT − VR. We also assessed the neurophysiologic basis putatively subtending the BR-based motor function recovery, using EEG recording during RAGT. Forty-five patients with stroke were enrolled in this study and randomized with a 1:2 ratio into either the RAGT + VR (n = 30) or the RAGT − VR (n = 15) group. The former group carried out rehabilitation training with the Lokomat©Pro; whereas, the latter used the Lokomat©Nanos. The rehabilitation protocol consisted of 40 one-hour training sessions. At the end of the training, the RAGT + VR improved in FMA-LE (p < 0.001) and BR (as per BES, (p < 0.001), and BUT, (p < 0.001)) more than the RAGT- did (p < 0.001). These differences in clinical outcomes were paralleled by a greater strengthening of visuomotor connectivity and corticomotor excitability (as detected at the EEG analyses) in the RAGT + VR than in the RAGT − VR (all comparisons p < 0.001), corresponding to an improved motor programming and execution in the former group.
We may argue that BR recovery was important concerning functional motor improvement by its integration with the motor control system. This likely occurred through the activation of the Mirror Neuron System secondary to the visuomotor feedback provision, resembling virtual reality. Last, our data further confirm the important role of visuomotor feedback in post-stroke rehabilitation, which can achieve better patient-tailored improvement in functional gait by means of RAGT + VR targeting BR.
... EEG analysis consisted of the computation of the power spectral density (PSD) and the temporal frequency (TF) analysis to evaluate Event-related-spectral perturbations. Spectrum analysis was carried using a standard fast Fourier transform (FFT) algorithm within ϑ (4-7 Hz), µ (8-12 Hz), β (12-30 Hz), low-γ (Lγ) (31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45), and high-γ (Hγ) (46-70 Hz) bands and related to the phases of the gait cycle [41]. After an intervention period lasting 2 months, the experimental group underwent Lokomat rehabilitation with VR showing higher motor recovery of the lower limbs with respect to patients not using VR technology. ...
... EEG analysis consisted of the computation of the power spectral density (PSD) and the temporal frequency (TF) analysis to evaluate Event-related-spectral perturbations. Spectrum analysis was carried using a standard fast Fourier transform (FFT) algorithm within ϑ (4-7 Hz), µ (8-12 Hz), β (12-30 Hz), low-γ (Lγ) (31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45), and high-γ (Hγ) (46-70 Hz) bands and related to the phases of the gait cycle [41]. After an intervention period lasting 2 months, the experimental group underwent Lokomat rehabilitation with VR showing higher motor recovery of the lower limbs with respect to patients not using VR technology. ...
Here we reviewed the last evidence on the application of electroencephalography (EEG) as a non-invasive and portable neuroimaging method useful to extract hallmarks of neuroplasticity induced by virtual reality (VR) rehabilitation approaches in stroke patients. In the neurorehabilitation context, VR training has been used extensively to hamper the effects of motor treatments on the stroke’s brain. The concept underlying VR therapy is to improve brain plasticity by engaging users in multisensory training. In this narrative review, we present the key concepts of VR protocols applied to the rehabilitation of stroke patients and critically discuss challenges of EEG signal when applied as endophenotype to extract neurophysiological markers. When VR technology was applied to magnify the effects of treatments on motor recovery, significant EEG-related neural improvements were detected in the primary motor circuit either in terms of power spectral density or as time-frequency domains.
