Shixiong Chen

Towards reducing the impacts of unwanted movements on identification of motion intentions

Surface electromyogram (sEMG) has been extensively used as a control signal in prosthesis devices. However, it is still a great challenge to make multifunctional myoelectric prostheses clinically available due to a number of critical issues associated with existing EMG based control strategy. One such issue would be the effect of unwanted movements (UMs) that are inadvertently done by users on the performance of movement classification in EMG pattern recognition based algorithms. Since UMs are not considered in training a classifier, they would decay the performance of a trained classifier in identifying the target movements (TMs), which would cause some undesired actions in control of multifunctional prostheses. In this study, the impact of UMs was systemically investigated in both able-bodied subjects and transradial amputees. Our results showed that the UMs would be unevenly classified into all classes of the TMs. To reduce the impact of the UMs on the performance of a classifier, a new training strategy that would categorize all possible UMs into a new movement class was proposed and a metric called Reject Ratio that is a measure of how many UMs is rejected by a trained classifier was adopted. The results showed that the average Reject Ratio across all the participants was greater than 91%, meanwhile the average classification accuracy of TMs was above 99% when UMs occurred. This suggests that the proposed training strategy could greatly reduce the impact of UMs on the performance of the trained classifier in identifying the TMs and may enhance the robustness of myoelectric control in clinical applications.

Stimulus frequency otoacoustic emissions evoked by swept tones

Otoacoustic emissions (OAEs) are soft sounds generated by the cochlea and the measurements of OAEs are useful in detecting cochlear damages. Stimulus frequency otoacoustic emissions (SFOAEs) are evoked by one single tone and they are the most frequency specific in probing functional status of the cochlea than other types of OAEs. However, SFOAEs are currently restricted to research only because of the difficulty and low efficiency of their measurements. To solve these problems, an efficient method of using swept tones to measure SFOAEs was proposed in this study. The swept tones had time-varying frequencies and therefore could efficiently measure SFOAEs over a wide frequency range with a resolution dependent on the sweep rate. A three-interval paradigm and a tracking filter were used to separate the swept-tone SFOAEs from background noises. The reliability of the swept-tone SFOAEs was examined by a repeated-measure design, and the accuracy was evaluated by the comparison with a standard method using pure tones as the stimuli. The pilot results of this study showed that SFOAEs could be measured successfully using swept tones in human ears with normal hearing. The amplitude and phase of the swept-tone SFOAEs were highly reproducible in the repeated measures, and were nearly equivalent to SFOAEs evoked by pure tones under various signal conditions. These findings suggest that the proposed swept-tone SFOAEs could be a useful method in estimating the cochlear functions and developing an efficient approach of OAE measurements to help with accurate hearing diagnoses in the clinic.

The development of biosensor with imaging ellipsometry

The concept of biosensor with imaging ellipsometry was proposed about ten years ago. It has become an automatic analysis technique for protein detection with merits of label-free, multi-protein analysis, and real-time analysis for protein interaction process, etc. Its principle, and related technique units, such as micro-array, micro-fluidic and bio-molecule interaction cell, sampling unit and calibration for quantitative detection as well as its applications in biomedicine field are presented here.

Bioelectric signal detrending using smoothness prior approach.

Bioelectric signals such as electromyogram (EMG) and electrocardiogram (ECG) are often affected by various low-frequency trending interferences. It is critical to remove these interferences from the recordings so that the critical features of the bioelectric signals could be clearly observed. In this study, an advanced method based on smoothness prior approach (SPA) was proposed to solve this problem. EMG and ECG signals from both the MIT-BIH database and the experiments were employed to evaluate the detrending performance of the proposed method. For comparison purposes, a conventional high-pass Butterworth filter was also used for the detrending of the EMG and ECG signals. Two numerical measures, the correlation coefficient (CC) and root mean square error (RMSE) between the clean data and the detrended data, were calculated to evaluate the detrending performance. The results showed that the proposed SPA method outperformed the high-pass filtering method in reducing various kinds of trending interferences and preserving the desired frequency contents of the EMG and ECG signals. The study suggested that the SPA method might be a promising approach in detrending bioelectric signals.

Effects of non-training movements on the performance of motion classification in electromyography pattern recognition

In electromyography pattern-recognition-based control of a multifunctional prosthesis, it would be inevitable for the users to unintentionally perform some classes of movements that are excluded from the training motion classes of a classifier, which might decay the performance of a trained classifier. It remains unknown how these untrained movements, designated as non-target movements (NTMs) in the study, would affect the performance of a trained classifier in the control of multifunctional prostheses. The goal of the current study was to evaluate the effects of NTMs on the performance of movement classification. Five classes of target movements (TMs) and four classes of NTMs were considered in this pilot study. A classifier based on a linear discriminant analysis (LDA) was trained with the electromyography (EMG) signals from the five TMs and the effects of the four NTMs were examined by feeding the EMG signals of the four NTMs to the trained classifier. Our results showed that these NTMs were classified into one or more classes of the TMs, which would cause the unexpected movements of prostheses. A method to reduce the effects of NTMs has been proposed in the study and our results showed that the averaged classification accuracies of the corrected classifiers were above 99% for the healthy subjects.

Activity Recognition Based on Pattern Recognition of Myoelectric Signals for Rehabilitation

Limb-amputation, stroke, trauma, and some other congenital anomalies not only decrease patients’ quality of life but also cause severe psychological burdens to them. Several advanced rehabilitation technologies have been developed to help patients with limb disabilities restore their lost motor functions. As a kind of neural signal, surface electromyogram (sEMG) recorded on limb muscles usually contain rich information associated with limb motions. By decoding the sEMG with pattern recognition techniques, the motion intents can be effectively identified and used for the control of rehabilitation devices. In this chapter, the control of upper-limb prostheses and rehabilitation robots based on the pattern recognition of sEMG signals was detailedly introduced and discussed. In addition, the clinical feasibility of sEMG-based pattern recognition technique towards an improved function restoration for upper-limb amputees and stroke survivors is also described.

Towards Real-Time Detection of Gait Events on Different Terrains Using Time-Frequency Analysis and Peak Heuristics Algorithm

Real-time detection of gait events can be applied as a reliable input to control drop foot correction devices and lower-limb prostheses. Among the different sensors used to acquire the signals associated with walking for gait event detection, the accelerometer is considered as a preferable sensor due to its convenience of use, small size, low cost, reliability, and low power consumption. Based on the acceleration signals, different algorithms have been proposed to detect toe off (TO) and heel strike (HS) gait events in previous studies. While these algorithms could achieve a relatively reasonable performance in gait event detection, they suffer from limitations such as poor real-time performance and are less reliable in the cases of up stair and down stair terrains. In this study, a new algorithm is proposed to detect the gait events on three walking terrains in real-time based on the analysis of acceleration jerk signals with a time-frequency method to obtain gait parameters, and then the determination of the peaks of jerk signals using peak heuristics. The performance of the newly proposed algorithm was evaluated with eight healthy subjects when they were walking on level ground, up stairs, and down stairs. Our experimental results showed that the mean F1 scores of the proposed algorithm were above 0.98 for HS event detection and 0.95 for TO event detection on the three terrains. This indicates that the current algorithm would be robust and accurate for gait event detection on different terrains. Findings from the current study suggest that the proposed method may be a preferable option in some applications such as drop foot correction devices and leg prostheses.

A new approach to mitigate the effect of force variation on pattern recognition for myoelectric control

In myoelectric prosthetic control, the motion classification performance would be decayed if an electromyography (EMG) pattern to be recognized differs significantly from the one used for classifier training. Generally, the training signals are acquired when a subject performs motions with a proper force. In practical use of a myoelectric prosthesis, however, the variation of force levels to do a motion would be inevitable, which will cause a change of EMG patterns. In this study, we proposed a Parallel Classification Strategy consisting of three parallel classifiers created at low, medium, and high force levels, respectively, and designed a regulation to categorize the input signals into corresponding classifiers for pattern recognition. The pilot experimental results demonstrated that the proposed method could enhance the classification accuracy at different force levels with an average classification rate of 98.8%, which was higher than the current method (91.9%). Thus, the proposed method might improve the control performance for present myoelectric prostheses.In myoelectric prosthetic control, the motion classification performance would be decayed if an electromyography (EMG) pattern to be recognized differs significantly from the one used for classifier training. Generally, the training signals are acquired when a subject performs motions with a proper force. In practical use of a myoelectric prosthesis, however, the variation of force levels to do a motion would be inevitable, which will cause a change of EMG patterns. In this study, we proposed a Parallel Classification Strategy consisting of three parallel classifiers created at low, medium, and high force levels, respectively, and designed a regulation to categorize the input signals into corresponding classifiers for pattern recognition. The pilot experimental results demonstrated that the proposed method could enhance the classification accuracy at different force levels with an average classification rate of 98.8%, which was higher than the current method (91.9%). Thus, the proposed method might improve the control performance for present myoelectric prostheses.

Muscle Activity-Driven Green-Oriented Random Number Generation Mechanism to Secure WBSN Wearable Device Communications

Wireless body sensor networks (WBSNs) mostly consist of low-cost sensor nodes and implanted devices which generally have extremely limited capability of computations and energy capabilities. Hence, traditional security protocols and privacy enhancing technologies are not applicable to the WBSNs since their computations and cryptographic primitives are normally exceedingly complicated. Nowadays, mobile wearable and wireless muscle-computer interfaces have been integrated with the WBSN sensors for various applications such as rehabilitation, sports, entertainment, and healthcare. In this paper, we propose MGRNG, a novel muscle activity-driven green-oriented random number generation mechanism which uses the human muscle activity as green energy resource to generate random numbers (RNs). The RNs can be used to enhance the privacy of wearable device communications and secure WBSNs for rehabilitation purposes. The method was tested on 10 healthy subjects as well as 5 amputee subjects with 105 segments of simultaneously recorded surface electromyography signals from their forearm muscles. The proposed MGRNG requires only one second to generate a 128-bit RN, which is much more efficient when compared to the electrocardiography-based RN generation algorithms. Experimental results show that the RNs generated from human muscle activity signals can pass the entropy test and the NIST random test and thus can be used to secure the WBSN nodes.

Evaluation of normal swallowing functions by using dynamic high-density surface electromyography maps

BackgroundSwallowing is a continuous process with substantive interdependencies among different muscles, and it plays a significant role in our daily life. The aim of this study was to propose a novel technique based on high-density surface electromyography (HD sEMG) for the evaluation of normal swallowing functions.MethodsA total of 96 electrodes were placed on the front neck to acquire myoelectric signals from 12 healthy subjects while they were performing different swallowing tasks. HD sEMG energy maps were constructed based on the root mean square values to visualize muscular activities during swallowing. The effects of different volumes, viscosities, and head postures on the normal swallowing process were systemically investigated by using the energy maps.ResultsThe results showed that the HD sEMG energy maps could provide detailed spatial and temporal properties of the muscle electrical activity, and visualize the muscle contractions that closely related to the swallowing function. The energy maps also showed that the swallowing time and effort was also explicitly affected by the volume and viscosity of the bolus. The concentration of the muscular activities shifted to the opposite side when the subjects turned their head to either side.ConclusionsThe proposed method could provide an alternative method to physiologically evaluate the dynamic characteristics of normal swallowing and had the advantage of providing a full picture of how different muscle activities cooperate in time and location. The findings from this study suggested that the HD sEMG technique might be a useful tool for fast screening and objective assessment of swallowing disorders or dysphagia.