Amputee Subjects Research Articles

Transradial Amputee Gesture Classification Using an Optimal Number of sEMG Sensors: An Approach Using ICA Clustering.

Surface electromyography (sEMG)-based pattern recognition studies have been widely used to improve the classification accuracy of upper limb gestures. Information extracted from multiple sensors of the sEMG recording sites can be used as inputs to control powered upper limb prostheses. However, usage of multiple EMG sensors on the prosthetic hand is not practical and makes it difficult for amputees due to electrode shift/movement, and often amputees feel discomfort in wearing sEMG sensor array. Instead, using fewer numbers of sensors would greatly improve the controllability of prosthetic devices and it would add dexterity and flexibility in their operation. In this paper, we propose a novel myoelectric control technique for identification of various gestures using the minimum number of sensors based on independent component analysis (ICA) and Icasso clustering. The proposed method is a model-based approach where a combination of source separation and Icasso clustering was utilized to improve the classification performance of independent finger movements for transradial amputee subjects. Two sEMG sensor combinations were investigated based on the muscle morphology and Icasso clustering and compared to Sequential Forward Selection (SFS) and greedy search algorithm. The performance of the proposed method has been validated with five transradial amputees, which reports a higher classification accuracy ( > 95%). The outcome of this study encourages possible extension of the proposed approach to real time prosthetic applications.

Foot-flat period estimation during daily living situations of asymptomatic and lower limb amputee subjects

his work was supported by the French National Research Agency [grant number ANR-292 2010-TECS-020].

Real-time simultaneous myoelectric control by transradial amputees using linear and probability-weighted regression.

Regression-based prosthesis control using surface electromyography (EMG) has demonstrated real-time simultaneous control of multiple degrees of freedom (DOFs) in transradial amputees. However, these systems have been limited to control of wrist DOFs. Use of intramuscular EMG has shown promise for both wrist and hand control in able-bodied subjects, but to date has not been evaluated in amputee subjects. The objective of this study was to evaluate two regression-based simultaneous control methods using intramuscular EMG in transradial amputees and compare their performance to able-bodied subjects. Two transradial amputees and sixteen able-bodied subjects used fine wire EMG recorded from six forearm muscles to control three wrist/hand DOFs: wrist rotation, wrist flexion/extension, and hand open/close. Both linear regression and probability-weighted regression systems were evaluated in a virtual Fitts' Law test. Though both amputee subjects initially produced worse performance metrics than the able-bodied subjects, the amputee subject who completed multiple experimental blocks of the Fitts' law task demonstrated substantial learning. This subject's performance was within the range of able-bodied subjects by the end of the experiment. Both amputee subjects also showed improved performance when using probability-weighted regression for targets requiring use of only one DOF, and mirrored statistically significant differences observed with able-bodied subjects. These results indicate that amputee subjects may require more learning to achieve similar performance metrics as able-bodied subjects. These results also demonstrate that comparative findings between linear and probability-weighted regression with able-bodied subjects reflect performance differences when used by the amputee population.

Combining human volitional control with intrinsic controller on robotic prosthesis: A case study on adaptive slope walking.

Affording lower-limb amputees the ability to volitionally control robotic prostheses can improve the adaptability to terrain changes as well as enhancing proprioception. However, it also increases amputees' conscious burdens for prosthesis control. Therefore, in this paper, we aim to propose a hybrid controller which combines human volitional control with the intrinsic controller on the robotic transtibial prosthesis, enabling the amputee actively controlling prosthesis with little conscious attention. In this preliminary study, a hybrid controller for adaptive slope walking was designed. A slope estimator was embedded in the intrinsic controller to estimate the ground slope of the previous step using signals measured by prosthetic sensors. And a myoelectric controller allows the amputee subject to convey slope changes to prosthetic controller by volitionally contract his residual muscles, whose electromyography signals were mapped to the slope increment. The hybrid controller combined these two results to obtain the estimated slope. One male transtibial amputee subject was recruited in this research. Experiment results showed that the intrinsic slope estimator produced satisfactory estimation results with an average absolute error of 0.70 ± 0.54 degrees. By adding amputee's volitional control, the hybrid controller is able to predict the upcoming slope changes.

A sensory feedback system for prosthetic hand based on evoked tactile sensation.

The lack of reliable sensory feedback has been one of the barriers in prosthetic hand development. Restoring sensory function from prosthetic hand to amputee remains a great challenge to neural engineering. In this paper, we present the development of a sensory feedback system based on the phenomenon of evoked tactile sensation (ETS) at the stump skin of residual limb induced by transcutaneous electrical nerve stimulation (TENS). The system could map a dynamic pattern of stimuli to an electrode placed on the corresponding projected finger areas on the stump skin. A pressure transducer placed at the tip of prosthetic fingers was used to sense contact pressure, and a high performance DSP processor sampled pressure signals, and calculated the amplitude of feedback stimulation in real-time. Biphasic and charge-balanced current pulses with amplitude modulation generated by a multi-channel laboratory stimulator were delivered to activate sensory nerves beneath the skin. We tested this sensory feedback system in amputee subjects. Preliminary results showed that the subjects could perceive different levels of pressure at the tip of prosthetic finger through evoked tactile sensation (ETS) with distinct grades and modalities. We demonstrated the feasibility to restore the perceptual sensation from prosthetic fingers to amputee based on the phenomenon of evoked tactile sensation (ETS) with TENS.

Residuals of autoregressive model providing additional information for feature extraction of pattern recognition-based myoelectric control.

Myoelectric control based on pattern recognition has been studied for several decades. Autoregressive (AR) features are one of the mostly used feature extraction methods among myoelectric control studies. Almost all previous studies only used the AR coefficients without the residuals of AR model for classification. However, the residuals of AR model contain important amplitude information of the electromyography (EMG) signals. In this study, we added the residuals to the AR features (AR+re) and compared its performance with the classical sixth-order AR coefficients. We tested six unilateral transradial amputees and eight able-bodied subjects for eleven hand and wrist motions. The classification accuracy (CA) of the intact side for amputee subjects and the right hand for able-bodied subjects showed that the CA of AR+re features was slightly but significantly higher than that of classical AR features (p = 0.009), which meant that residuals could provide additional information to classical AR features for classification. Interestingly, the CA of the affected side for amputee subjects showed that there was no significant difference between the CA of AR+re features and classical AR features (p > 0.05). We attributed this to the fact that the amputee subjects could not use their affected side to produce consistent EMG patterns as their intact side or the dominant hand of the able-bodied subjects. Since the residuals were already available when the AR coefficients were computed, the results of this study suggested adding the residuals to classical AR features to potentially improve the performance of pattern recognition-based myoelectric control.

Enhanced robustness of myoelectric pattern recognition to across-day variation through invariant feature extraction.

Robust pattern recognition is critical for myoelectric prosthesis (MP) developed in the laboratory to be used in real life. This study focuses on the robustness of MP control during the usage across many days. Due to the variability inhered in extended electromyography (EMG) signals, the distribution of EMG features extracted from several days' data may have large intra-class scatter. However, as the subjects perform the same motion type in different days, we hypothesize there exist some invariant characteristics in the EMG features. Therefore, give a set of training data from several days, it is possible to find an invariant component in them. To this end, an invariant feature extraction (IFE) framework based on kernel fisher discriminant analysis is proposed. A desired transformation, which minimizes the intra-class (within a motion type) scatter meanwhile maximizes the inter-class (between different motion types) scatter, is found. Five intact-limbed subjects and three transradial-amputee subjects participated in an experiment lasting ten days. The results show that the generalization ability of the classifier trained on previous days to the unseen testing days can be improved by IFE. IFE significantly outperforms Baseline (original input feature) in classification accuracy, both for intact-limbed subjects and amputee subjects (average 88.97% vs. 91.20% and 85.09% vs. 88.22%, p <; 0.05).

Metabolics of stair ascent with a powered transfemoral prosthesis.

This paper evaluates the effectiveness of a powered knee and ankle prosthesis for stair ascent through a metabolic assessment comparing energy expenditure of a single transfemoral amputee subject while ascending stairs with the powered prosthesis relative to his passive daily use device, as well as comparing the kinematics and kinetics obtained with the passive prosthesis to healthy biomechanics. The subject wore a portable system that measured pulmonary gaseous exchange rates of oxygen and carbon dioxide while he ascended stairs with each of the prostheses in alternating tests. The results indicated that the amputee's energy expenditure decreased by 32 percent while climbing with the powered prosthesis as compared to his passive one, and the kinematics and kinetics achieved were representative of healthy biomechanics.

Investigation of Timing to Switch Control Mode in Powered Knee Prostheses during Task Transitions.

Current powered prosthetic legs require switching control modes according to the task the user is performing (e.g. level-ground walking, stair climbing, walking on slopes, etc.). To allow prosthesis users safely and seamlessly transition between tasks, it is critical to determine when to switch the prosthesis control mode during task transitions. Our previous study defined critical timings for different types of task transitions in ambulation; however, it is unknown whether it is the unique timing that allows safe and seamless transitions. The goals of this study were to (1) systematically investigate the effects of mode switch timing on the prosthesis user’s performance in task transitions, and (2) identify appropriate timing to switch the prosthesis control mode so that the users can seamlessly transition between different locomotion tasks. Five able-bodied (AB) and two transfemoral (TF) amputee subjects were tested as they wore a powered knee prosthesis. The prosthesis control mode was switched manually at various times while the subjects performed different types of task transitions. The subjects’ task transition performances were evaluated by their walking balance and success in performing seamless task transitions. The results demonstrated that there existed a time window within which switching the prosthesis control mode neither interrupted the subjects’ task transitions nor disturbed their walking balance. Therefore, the results suggested the control mode switching of a lower limb prosthesis can be triggered within an appropriate time window instead of a specific timing or an individual phase. In addition, a generalized criterion to determine the appropriate mode switch timing was proposed. The outcomes of this study could provide important guidance for future designs of neurally controlled powered knee prostheses that are safe and reliable to use.

Context-Dependent Upper Limb Prosthesis Control for Natural and Robust Use.

Pattern recognition and regression methods applied to the surface EMG have been used for estimating the user intended motor tasks across multiple degrees of freedom (DOF), for prosthetic control. While these methods are effective in several conditions, they are still characterized by some shortcomings. In this study we propose a methodology that combines these two approaches for mutually alleviating their limitations. This resulted in a control method capable of context-dependent movement estimation that switched automatically between sequential (one DOF at a time) or simultaneous (multiple DOF) prosthesis control, based on an online estimation of signal dimensionality. The proposed method was evaluated in scenarios close to real-life situations, with the control of a physical prosthesis in applied tasks of varying difficulties. Test prostheses were individually manufactured for both able-bodied and transradial amputee subjects. With these prostheses, two amputees performed the Southampton Hand Assessment Procedure test with scores of 58 and 71 points. The five able-bodied individuals performed standardized tests, such as the box&block and clothes pin test, reducing the completion times by up to 30%, with respect to using a state-of-the-art pure sequential control algorithm. Apart from facilitating fast simultaneous movements, the proposed control scheme was also more intuitive to use, since human movements are predominated by simultaneous activations across joints. The proposed method thus represents a significant step towards intelligent, intuitive and natural control of upper limb prostheses.

Actuator With Angle-Dependent Elasticity for Biomimetic Transfemoral Prostheses

Despite tremendous improvements in recent years, lower-limb prostheses are still inferior to their biological counterparts. Most powered knee joints use impedance control, but it is unknown which impedance profiles are needed to replicate physiological behavior. Recently, we have developed a method to quantify such profiles from conventional gait data. Based on this method, we derive stiffness requirements for knee prostheses, and we propose an actuation concept where physical actuator stiffness changes in function of joint angle. The idea is to express stiffness and moment requirements as functions of angle, and then to combine a series elastic actuator (SEA) with an optimized nonlinear transmission and parallel springs to reproduce the profiles. By considering the angle-dependent stiffness requirement, the upper bound for the impedance in zero-force control could be reduced by a factor of two. We realize this ANGle-dependent ELAstic Actuator (ANGELAA) in a leg, with rubber cords as series elastic elements. Hysteresis in the rubber is accounted for, and knee moment is estimated with a mean error of 0.7 Nm. The nonlinear parallel elasticity creates equilibria near 0 $^\circ$ as well as 90 $^\circ$ knee flexion, frequent postures in daily life. Experimental evaluation in a test setup shows force control bandwidth around 5–9 Hz, and a pilot experiment with an amputee subject shows the feasibility of the approach. While weight and power consumption are not optimized in this prototype, the incorporated mechatronic principles may pave the way for cheaper and lighter actuators in artificial legs and in other applications where stiffness requirements depend on kinematic configuration.

Towards Zero Retraining for Myoelectric Control Based on Common Model Component Analysis.

In spite of several decades of intense research and development, the existing algorithms of myoelectric pattern recognition (MPR) are yet to satisfy the criteria that a practical upper extremity prostheses should fulfill. This study focuses on the criterion of the short, or even zero subject training. Due to the inherent nonstationarity in surface electromyography (sEMG) signals, current myoelectric control algorithms usually need to be retrained daily during a multiple days' usage. This study was conducted based on the hypothesis that there exist some invariant characteristics in the sEMG signals when a subject performs the same motion in different days. Therefore, given a set of classifiers (models) trained on several days, it is possible to find common characteristics among them. To this end, we proposed to use common model component analysis (CMCA) framework, in which an optimized projection was found to minimize the dissimilarity among multiple models of linear discriminant analysis (LDA) trained using data from different days. Five intact-limbed subjects and two transradial amputee subjects participated in an experiment including six sessions of sEMG data recording, which were performed in six different days, to simulate the application of MPR over multiple days. The results demonstrate that CMCA has a significant better generalization ability with unseen data (not included in the training data), leading to classification accuracy improvement and increase of completion rate in a motion test simulation, when comparing with the baseline reference method. The results indicate that CMCA holds a great potential in the effort of developing zero retraining of MPR.

Fuzzy-Logic-Based Terrain Identification with Multisensor Fusion for Transtibial Amputees

Terrain identification is essential for the control of robotic transtibial prostheses to realize smooth locomotion transitions. In this paper, we present a real-time fuzzy-logic-based terrain identification method with multisensor fusion. Five locomotion features, including the foot inclination angle at the first strike, the shank inclination angle at the first strike, foot strike sequence, the foot inclination angle at mid-stance, and the shank inclination angle at toe-off, are used to identify different terrains and terrain transitions. These features are measured by the fusion of two triaxis gyroscopes, two triaxis accelerometers, two force sensitive resistors, and a timer, which can be embedded into the prosthesis. Based on these features, a fuzzy-logic-based identification method is proposed to identify five terrains: level ground, stair ascent, stair descent, ramp ascent, and ramp descent. Moreover, a transition constraint function is developed to improve the identification performance. The execution time of the identification method is 0.79 ms ± 0.02 ms (mean ± standard error of mean) and continuous terrain identification results show that the method can be operated online in real time. The average identification accuracy of 98.74% ± 0.32% is obtained from experiments with six able-bodied and three amputee subjects during steady locomotion periods (no terrain transition). In locomotion transition periods, all the eight transitions we studied are correctly identified and the average identification delay is 9.06% ± 3.46% of one gait cycle.

Detection of Prosthetic Knee Movement Phases via In-Socket Sensors: A Feasibility Study.

This paper presents an approach of identifying prosthetic knee movements through pattern recognition of mechanical responses at the internal socket's wall. A quadrilateral double socket was custom made and instrumented with two force sensing resistors (FSR) attached to specific anterior and posterior sites of the socket's wall. A second setup was established by attaching three piezoelectric sensors at the anterior distal, anterior proximal, and posterior sites. Gait cycle and locomotion movements such as stair ascent and sit to stand were adopted to characterize the validity of the technique. FSR and piezoelectric outputs were measured with reference to the knee angle during each phase. Piezoelectric sensors could identify the movement of midswing and terminal swing, pre-full standing, pull-up at gait, sit to stand, and stair ascent. In contrast, FSR could estimate the gait cycle stance and swing phases and identify the pre-full standing at sit to stand. FSR showed less variation during sit to stand and stair ascent to sensitively represent the different movement states. The study highlighted the capacity of using in-socket sensors for knee movement identification. In addition, it validated the efficacy of the system and warrants further investigation with more amputee subjects and different sockets types.

Ankle–knee prosthesis with active ankle and energy transfer: Development of the CYBERLEGs Alpha-Prosthesis

This paper presents the development of the CYBERLEGs Alpha-Prototype prosthesis, a new transfemoral prosthesis incorporating a new variable stiffness ankle actuator based on the MACCEPA architecture, a passive knee with two locking mechanisms, and an energy transfer mechanism that harvests negative work from the knee and delivers it to the ankle to assist pushoff. The CYBERLEGs Alpha-Prosthesis is part of the CYBERLEGs FP7-ICT project, which combines a prosthesis system to replace a lost limb in parallel with an exoskeleton to assist the sound leg, and sensory array to control both systems. The prosthesis attempts to produce a natural level ground walking gait that approximates the joint torques and kinematics of a non-amputee while maintaining compliant joints, which has the potential to decrease impulsive losses, and ultimately reduce the end user energy consumption. This first prototype consists of a passive knee and an active ankle which are energetically coupled to reduce the total power consumption of the device. Here we present simulations of the actuation system of the ankle and the passive behavior of the knee module with and without the energy transfer effects, the mechanical design of the prosthesis, and empirical results from testing of the physical device with amputee subjects.

Piezoelectric bimorphs' characteristics as in-socket sensors for transfemoral amputees.

Alternative sensory systems for the development of prosthetic knees are being increasingly highlighted nowadays, due to the rapid advancements in the field of lower limb prosthetics. This study presents the use of piezoelectric bimorphs as in-socket sensors for transfemoral amputees. An Instron machine was used in the calibration procedure and the corresponding output data were further analyzed to determine the static and dynamic characteristics of the piezoelectric bimorph. The piezoelectric bimorph showed appropriate static operating range, repeatability, hysteresis, and frequency response for application in lower prosthesis, with a force range of 0–100 N. To further validate this finding, an experiment was conducted with a single transfemoral amputee subject to measure the stump/socket pressure using the piezoelectric bimorph embedded inside the socket. The results showed that a maximum interface pressure of about 27 kPa occurred at the anterior proximal site compared to the anterior distal and posterior sites, consistent with values published in other studies. This paper highlighted the capacity of piezoelectric bimorphs to perform as in-socket sensors for transfemoral amputees. However, further experiments are recommended to be conducted with different amputees with different socket types.

Virtual Constraint Control of a Powered Prosthetic Leg: From Simulation to Experiments With Transfemoral Amputees

Recent powered (or robotic) prosthetic legs independently control different joints and time periods of the gait cycle, resulting in control parameters and switching rules that can be difficult to tune by clinicians. This challenge might be addressed by a unifying control model used by recent bipedal robots, in which virtual constraints define joint patterns as functions of a monotonic variable that continuously represents the gait cycle phase. In the first application of virtual constraints to amputee locomotion, this paper derives exact and approximate control laws for a partial feedback linearization to enforce virtual constraints on a prosthetic leg. We then encode a human-inspired invariance property called effective shape into virtual constraints for the stance period. After simulating the robustness of the partial feedback linearization to clinically meaningful conditions, we experimentally implement this control strategy on a powered transfemoral leg. We report the results of three amputee subjects walking overground and at variable cadences on a treadmill, demonstrating the clinical viability of this novel control approach.

A Robotic Leg Prosthesis: Design, Control, and Implementation

This article describes the design and control of a powered knee and ankle prosthesis for transfemoral amputees. Following a description of the design hardware, a hybrid control approach that provides coordination for level walking is described. The hybrid control approach combines a piecewise-passive impedance-based component during the stance phase of gait with a high impedance trajectory-tracking component during the terminal stance and swing. To validate the design, the controller was implemented on the powered prosthesis prototype, and its ability to provide level walking functionality was evaluated on three transfemoral amputee subjects. The data presented from these experimental trials indicate that the prosthesis and control approach reproduce knee and ankle joint kinematic and kinetic features that are highly representative of corresponding healthy joint biomechanics.

A functional model and simulation of spinal motor pools and intrafascicular recordings of motoneuron activity in peripheral nerve.

Decoding motor intent from recorded neural signals is essential for the development of effective neural-controlled prostheses. To facilitate the development of online decoding algorithms we have developed a software platform to simulate neural motor signals recorded with peripheral nerve electrodes, such as longitudinal intrafascicular electrodes (LIFEs). The simulator uses stored motor intent signals to drive a pool of simulated motoneurons with various spike shapes, recruitment characteristics, and firing frequencies. Each electrode records a weighted sum of a subset of simulated motoneuron activity patterns. As designed, the simulator facilitates development of a suite of test scenarios that would not be possible with actual data sets because, unlike with actual recordings, in the simulator the individual contributions to the simulated composite recordings are known and can be methodically varied across a set of simulation runs. In this manner, the simulation tool is suitable for iterative development of real-time decoding algorithms prior to definitive evaluation in amputee subjects with implanted electrodes. The simulation tool was used to produce data sets that demonstrate its ability to capture some features of neural recordings that pose challenges for decoding algorithms.

Analysis of using EMG and mechanical sensors to enhance intent recognition in powered lower limb prostheses

Objective. The purpose of this study was to determine the contribution of electromyography (EMG) data, in combination with a diverse array of mechanical sensors, to locomotion mode intent recognition in transfemoral amputees using powered prostheses. Additionally, we determined the effect of adding time history information using a dynamic Bayesian network (DBN) for both the mechanical and EMG sensors. Approach. EMG signals from the residual limbs of amputees have been proposed to enhance pattern recognition‐based intent recognition systems for powered lower limb prostheses, but mechanical sensors on the prosthesis—such as inertial measurement units, position and velocity sensors, and load cells—may be just as useful. EMG and mechanical sensor data were collected from 8 transfemoral amputees using a powered knee/ankle prosthesis over basic locomotion modes such as walking, slopes and stairs. An offline study was conducted to determine the benefit of different sensor sets for predicting intent. Main results. EMG information was not as accurate alone as mechanical sensor information (p < 0.05) for any classification strategy. However, EMG in combination with the mechanical sensor data did significantly reduce intent recognition errors (p < 0.05) both for transitions between locomotion modes and steady-state locomotion. The sensor time history (DBN) classifier significantly reduced error rates compared to a linear discriminant classifier for steady-state steps, without increasing the transitional error, for both EMG and mechanical sensors. Combining EMG and mechanical sensor data with sensor time history reduced the average transitional error from 18.4% to 12.2% and the average steady-state error from 3.8% to 1.0% when classifying level-ground walking, ramps, and stairs in eight transfemoral amputee subjects. Significance. These results suggest that a neural interface in combination with time history methods for locomotion mode classification can enhance intent recognition performance; this strategy should be considered for future real-time experiments.