Increasing the robustness against force variation in EMG motion classification by common spatial patterns.

In the practical use of an electromyography (EMG) pattern-recognition based myoelectric prosthesis, the variation of force levels to do a motion would be inevitable, which will cause a change of EMG patterns. Therefore, the force variation will decay the performance of a trained classifier. In this study, the common spatial pattern (CSP) method was proposed with an attempt to improve the robustness of EMG-PR based classifier against force variation. The EMG signals were acquired from three able-bodied subjects when they were performing the motions at low, medium, and high force levels, respectively. And in the pattern recognition, CSP features were extracted from the EMG signals for motion classification. By comparing the classification accuracies between the CSP and the commonly used time-domain (TD) features, the CSP features showed a better robustness against force variation with an increment of 5.3% of the average classification accuracy. Especially, the classification accuracy of a classifier was 84.2% when tested at low force level by using CSP features, which was 18.5% higher than that of the TD features. These preliminary results suggest that using CSP features may increase the robustness of EMG-based myoelectric control.

Using spatial features for classification of combined motions based on common spatial pattern.

Motion recognition is an important application of electromyography (EMG) analysis. While discrete motions such as hand open, hand close and wrist pronation have been extensively investigated, studies on combined motions involving two or more degrees of freedom (DOFs) are relatively few and the classification accuracy of the combined motions reported in previous studies is barely satisfactory. To improve the accuracy of the combined motion recognition, common spatial pattern (CSP) was employed in this study to extract spatial features. 18 forearm motion classes, consisted of 8 discrete motions and 10 combined motions, were classified by the proposed method. Our results showed that the accuracy rate of CSP features was 96.3%, which outperformed the commonly used time-domain (TD) features by 2.4% and TD combined with auto-regression coefficients (TDAR) by 0.6%. Moreover, CSP features cost noticeable much less time than TDAR and quite less time than TD in testing. These results suggest that CSP features could be a better feature set for multi-DOF myoelectric control than conventional features.

Comparison of different shielding methods in acquisition of physiological signals.

Power line interference in the surrounding environment could usually introduce many difficulties when collecting and analyzing physiological signals. Since power line interference is usually several orders of amplitude larger than the physiological electrical signals, methods of suppressing power line interference should be considered during the signal acquisition. Many studies used a hardware or software band-stop filter to suppress power line interference but it could easily cause attenuations and distortions to the signal of interest. In this study, two kinds of methods that used different signals to drive the shields of the electrodes were proposed to reduce the impacts of power line interference. Three channels of two physiological signals (ECG and EMG) were simultaneously collected when the electrodes were not shielded (No-Shield), shielded by ground signals (GND-Shield) and shielded by buffered signals of the corresponding electrodes (Active-Shield), respectively, on a custom hardware platform based on TI ADS1299. The results showed that power line interference would be significantly suppressed when using shielding approaches, and the Active-Shield method could achieve the best performance with a power line interference reduction up to 36dB. The study suggested that the Active-Shield method at the analog front-end was a great candidate to reduce power line interference in routine acquisitions of physiological signals.

Angled tooth segmentation from computerized tomography images.

Tooth contour segmentation from dental computerized tomography (CT) images is one of the fundamental steps in reconstructing the three-dimensional models of teeth. However, existing methods depend on the tooth shape similarity between adjacent slices, and have difficulty to segment these angled teeth whose contour shapes from adjacent slices may differ a lot. This study proposes a new method for contour segmentation of angled teeth from CT images. The volume of interest (VOI) of target tooth and corresponding tooth axis are first extracted from volumetric CT images. Local images within the VOI of target tooth are then rotated such that the tooth axis in the rotated images is perpendicular to the transverse section. Tooth contours are finally segmented from the rotated images using a hybrid level set model slice-by-slice. Experimental results verified that the proposed method was effective to segment contours of angled teeth from CT images.

Assessment of elbow spasticity with surface electromyography and mechanomyography based on support vector machine.

The Modified Ashworth Scale (MAS) is the gold standard in clinical for grading spasticity. However, its results greatly depend on the physician evaluations and are subjective. In this study, we investigated the feasibility of using support vector machine (SVM) to objectively assess elbow spasticity based on both surface electromyography (sEMG) and mechanomyography (MMG). sEMG signals and tri-axial accelerometer mechanomyography (ACC-MMG) signals were recorded simultaneously on patients' biceps and triceps when they extended or bended elbow passively. 39 post-stroke patients participated in the study, and were divided into four groups regarding MAS level (MAS=0, 1, 1+ or 2). The three types of features, root mean square (RMS), mean power frequency (MPF), and median frequency (MF), were calculated from sEMG and MMG signal recordings. Spearman correlation analysis was used to investigate the relationship between the features and spasticity grades. The results showed that the correlation between MAS and each of the five features (MMG-RMS of the biceps, MMG-RMS of the triceps, the EMG-RMS of the biceps, EMG-RMS of the triceps, EMG-MPF of the triceps) was significant (p<;0.05). The four spasticity grades were identified with SVM, and the classification accuracy of SVM with sEMG, MMG, sEMG-MMG were 70.9%, 83.3%, 91.7%, respectively. Our results suggest that using the SVM-based method with sEMG and MMG to assess elbow spasticity would be suitable for clinical management of spasticity.

A wearable 12-lead ECG acquisition system with fabric electrodes.

Continuous electrocardiogram (ECG) monitoring is significant for prevention of heart disease and is becoming an important part of personal and family health care. In most of the existing wearable solutions, conventional metal sensors and corresponding chips are simply integrated into clothes and usually could only collect few leads of ECG signals that could not provide enough information for diagnosis of cardiac diseases such as arrhythmia and myocardial ischemia. In this study, a wearable 12-lead ECG acquisition system with fabric electrodes was developed and could simultaneously process 12 leads of ECG signals. By integrating the fabric electrodes into a T-shirt, the wearable system would provide a comfortable and convenient user interface for ECG recording. For comparison, the proposed fabric electrode and the gelled traditional metal electrodes were used to collect ECG signals on a subject, respectively. The approximate entropy (ApEn) of ECG signals from both types of electrodes were calculated. The experimental results show that the fabric electrodes could achieve similar performance as the gelled metal electrodes. This preliminary work has demonstrated that the developed ECG system with fabric electrodes could be utilized for wearable health management and telemedicine applications.

A canonical correlation analysis based EMG classification algorithm for eliminating electrode shift effect.

Motion classification system based on surface Electromyography (sEMG) pattern recognition has achieved good results in experimental condition. But it is still a challenge for clinical implement and practical application. Many factors contribute to the difficulty of clinical use of the EMG based dexterous control. The most obvious and important is the noise in the EMG signal caused by electrode shift, muscle fatigue, motion artifact, inherent instability of signal and biological signals such as Electrocardiogram. In this paper, a novel method based on Canonical Correlation Analysis (CCA) was developed to eliminate the reduction of classification accuracy caused by electrode shift. The average classification accuracy of our method were above 95% for the healthy subjects. In the process, we validated the influence of electrode shift on motion classification accuracy and discovered the strong correlation with correlation coefficient of >0.9 between shift position data and normal position data.

Molar axis estimation from computed tomography images.

Estimation of tooth axis is needed for some clinical dental treatment. Existing methods require to segment the tooth volume from Computed Tomography (CT) images, and then estimate the axis from the tooth volume. However, they may fail during estimating molar axis due to that the tooth segmentation from CT images is challenging and current segmentation methods may get poor segmentation results especially for these molars with angle which will result in the failure of axis estimation. To resolve this problem, this paper proposes a new method for molar axis estimation from CT images. The key innovation point is that: instead of estimating the 3D axis of each molar from the segmented volume, the method estimates the 3D axis from two projection images. The method includes three steps. (1) The 3D images of each molar are projected to two 2D image planes. (2) The molar contour are segmented and the contour's 2D axis are extracted in each 2D projection image. Principal Component Analysis (PCA) and a modified symmetry axis detection algorithm are employed to extract the 2D axis from the segmented molar contour. (3) A 3D molar axis is obtained by combining the two 2D axes. Experimental results verified that the proposed method was effective to estimate the axis of molar from CT images.

Orthodontic force simulation of Tooth-PDL-Bone Complex under archwire loading.

Orthodontic force simulation is very important for the guidance of clinical orthodontic treatment. Previous works were mainly conducted by directly loading the force on a single or a few teeth. However, in clinic, the orthodontic force is provided by loading orthodontic appliances, and there currently is no appropriate way to measure the force on the teeth provided by the loaded appliances. This study presents a method to simulate the orthodontic force of a whole Tooth-Periodontal Ligament-Bone Complex (TPBC) by directly loading the archwire to the dentition applying the finite element method. A 3D TPBC model was reconstructed from CT images, and models of brackets and the archwire were also built. The loading procedure of the archwire was implemented by simulating the deformation and displacement of the archwire before and after the loading, and the stress of the archwire induced by the deformation and displacement was obtained. Then, the stress was applied to the brackets, and the corresponding orthodontic force of the TPBC was simulated. By applying the method, archwire designed according to the planned dentition shape was loaded successfully to the original dentition, and the orthodontic force of the TPBC was obtained.

Effects of contralateral acoustic stimulation on otoacoustic emissions induced by swept tones.

The medial olivocochlear complex (MOC) is an auditory nucleus that projects efferent nerve fibers to control the behaviors of both sides of the cochlea. Otoacoutsic emissions (OAEs) are by-products the activities of the outer hair cells (OHCs) in the cochlea and could be used as a noninvasive way to study the efferent control of the MOC. However, existing results regarding the efferent control are quite controversial and often restricted to a rather limited frequency range. In this study, a new method of measuring stimulus frequency otoacoustic emissions (SFOAEs) with the presence of a contralateral acoustic stimulation (CAS) was proposed to study the efferent control over the cochlea. SFOAEs were measured with swept tones with time varying frequencies so that SFOAE spectra with and without the presence of the CAS could be compared with high frequency-resolution. The results showed that there was consistent decrease in the amplitude of the swept-tone SFOAEs across a wide frequency range from 0.5 to 8 kHz when the CAS was presented, suggesting an outstanding attenuation of OHC activities by the efferent control from the MOC. The SFOAE decrease with the presence of the CAS might provide a new approach to measure the strength of the efferent control and to evaluate the functional status of the central auditory pathway.