An optoelectronic muscle contraction sensor for prosthetic hand application.

Surface electromyography (sEMG) is considered an established means for controlling prosthetic devices. sEMG suffers from serious issues such as electrical noise, motion artifact, complex acquisition circuitry, and high measuring costs because of which other techniques have gained attention. This work presents a new optoelectronic muscle (OM) sensor setup as an alternative to the EMG sensor for precise measurement of muscle activity. The sensor integrates a near-infrared light-emitting diode and phototransistor pair along with the suitable driver circuitry. The sensor measures skin surface displacement (that occurs during muscle contraction) by detecting backscattered infrared light from skeletal muscle tissue. With an appropriate signal processing scheme, the sensor was able to produce a 0-5 V output proportional to the muscular contraction. The developed sensor depicted decent static and dynamic features. In detecting muscle contractions from the forearm muscles of subjects, the sensor showed good similarity with the EMG sensor. In addition, the sensor displayed higher signal-to-noise ratio values and better signal stability than the EMG sensor. Furthermore, the OM sensor setup was utilized to control the rotation of the servomotor using an appropriate control scheme. Hence, the developed sensing system can measure muscle contraction information for controlling assistive devices.

Detection of COVID-19 Infection in CT and X-ray images using transfer learning approach.

BACKGROUND: The infection caused by the SARS-CoV-2 (COVID-19) pandemic is a threat to human lives. An early and accurate diagnosis is necessary for treatment. OBJECTIVE: The study presents an efficient classification methodology for precise identification of infection caused by COVID-19 using CT and X-ray images. METHODS: The depthwise separable convolution-based model of MobileNet V2 was exploited for feature extraction. The features of infection were supplied to the SVM classifier for training which produced accurate classification results. RESULT: The accuracies for CT and X-ray images are 99.42% and 98.54% respectively. The MCC score was used to avoid any mislead caused by accuracy and F1 score as it is more mathematically balanced metric. The MCC scores obtained for CT and X-ray were 0.9852 and 0.9657, respectively. The Youden's index showed a significant improvement of more than 2% for both imaging techniques. CONCLUSION: The proposed transfer learning-based approach obtained the best results for all evaluation metrics and produced reliable results for the accurate identification of COVID-19 symptoms. This study can help in reducing the time in diagnosis of the infection.

A low-cost transradial prosthesis controlled by the intention of muscular contraction.

Persons with upper-limb amputations face severe problems due to a reduction in their ability to perform the activities of daily living. The prosthesis controlled by electromyography (EMG) or other signals from sensors, switches, accelerometers, etc., can somewhat regain the lost capability of such individuals. However, there are several issues with these prostheses, such as expensive cost, limited functionality, unnatural control, slow operating speed, complexity, heavyweight, large size, etc. This paper proposes an affordable transradial prosthesis, controlled by the muscular contractions from user intention. A surface EMG sensor was explicitly fabricated for capturing the muscle contraction information from the residual forearm of subjects with amputation. An under actuated 3D printed hand was developed with a prosthetic socket assembly to attach the remaining upper-limb of such subjects. The hand integrates an intuitive closed-loop control system that receives reference input from the designed sensor and feedback input from a force sensor installed at the thumb tip. The performance of the EMG sensor was compared with that of a traditional sensor in detecting muscle contractions from the subjects. The designed sensor showed a good correlation (r > 0.93) and a better signal-to-noise ratio (SNR) feature to the conventional sensor. Further, a successful trial of the developed hand prosthesis was made on five different subjects with transradial amputation. The users wearing the hand prototype were able to perform faster and delicate grasping of various objects. The implemented control system allowed the prosthesis users to control the grasp force of hand fingers with their intention of muscular contractions.

Human Locomotion Classification for Different Terrains Using Machine Learning Techniques.

Gait analysis on healthy subjects was performed based on surface electromyographic and acceleration sensor signal, implemented through machine learning approaches. The surface EMG and 3-axes acceleration signals have been acquired for 5 different terrains: level ground, ramp ascent, ramp descent, stair ascent, and stair descent. These signals were acquired from the tibialis anterior and gastrocnemius medial head muscles that correspond to dorsiflexion and plantar flexion, respectively. After feature extraction, these signals are fed to 5 conventional classifiers: linear discriminant analysis, k-nearest neighbors, decision tree, random forest, and support vector machine, that classify different terrains for human locomotion. We compared the classification results for the above classifiers with deep neural network classifier. The objective was to obtain the features and classifiers that are able to discriminate between 5 locomotion terrains with maximum classification accuracy in minimum time by acquiring the signal from the least number of leg muscles. The results indicated that the support vector machine gives the highest classification accuracy of 99.20 (± 0.80)% for the dataset acquired from 15 healthy subjects. In terms of both accuracy and computation time, the support vector machine outperforms other classifiers.

A compact-sized surface EMG sensor for myoelectric hand prosthesis.

Myoelectric prosthesis requires a sensor that can reliably capture surface electromyography (sEMG) signal from amputees for its controlled operation. The main problems with the presently available EMG devices are their extremely high cost, large response time, noise susceptibility, less amplitude sensitivity, and larger size. This paper proposes a compact and affordable EMG sensor for the prosthetic application. The sensor consists of an electrode interface, signal conditioning unit, and power supply unit all encased in a single package. The performance of dry electrodes employed in the skin interface was compared with the conventional Ag/AgCl electrodes, and the results were found satisfactory. The envelope detection technique in the sensor based on the tuned RC parameters enables the generation of smooth, faster, and repeatable EMG envelope irrespective of signal strength and subject variability. The output performance of the developed sensor was compared with commercial EMG sensor regarding signal-to-noise ratio, sensitivity, and response time. To perform this, EMG data with both devices were recorded for 10 subjects (3 amputees and 7 healthy subjects). The results showed 1.4 times greater SNR values and 45% higher sensitivity of the developed sensor than the commercial EMG sensor. Also, the proposed sensor was 57% faster than the commercial sensor in producing the output response. The sEMG sensor was further tested on amputees to control the operation of a self-designed 3D printed prosthetic hand. With proportional control scheme, the myoelectric hand setup was able to provide quicker and delicate grasping of objects as per the strength of the EMG signal.

A low-cost, wearable sEMG sensor for upper limb prosthetic application.

This paper proposes a low-cost and sensitive surface electromyography (sEMG) sensor for the myoelectric prosthesis. The sensor consists of a skin interface, signal conditioning circuitry and power supply unit all encased in a single package. The tuned RC parameters based envelope detection scheme employed in the sensor enables faster as well as reliable recognition of EMG signal patterns regardless of its strength and subject variability. The output performance of the developed sensor was compared with a commercial EMG sensor regarding signal-to-noise ratio (SNR), amplitude sensitivity and response time. EMG signals with both the devices were acquired for 10 subjects (three amputees and seven healthy subjects), to perform this comparative analysis. The results showed 4× greater SNR values and 50% higher sensitivity of the developed sensor than the commercial EMG sensor. Also, the proposed sensor was 57% faster than the commercial sensor in producing the output response. The sensor was successfully tested on amputees for controlling a 3D printed hand prototype utilising a proportional control strategy. The enhanced output parameters of the sensor were responsible for smooth, faster and intuitive actuation of the prosthetic hand fingers.

Feature selection using regularized neighbourhood component analysis to enhance the classification performance of motor imagery signals.

In motor imagery (MI) based brain-computer interface (BCI) signal analysis, mu and beta rhythms of electroencephalograms (EEGs) are widely investigated due to their high temporal resolution and capability to define the different movement-related mental tasks separately. However, due to the high dimensions and subject-specific behaviour of EEG features, there is a need for a suitable feature selection algorithm that can select the optimal features to give the best classification performance along with increased computational efficiency. The present study proposes a feature selection algorithm based on neighbourhood component analysis (NCA) with modification of the regularization parameter. In the experiment, time, frequency, and phase features of the EEG are extracted using a dual-tree complex wavelet transform (DTCWT). Afterwards, the proposed algorithm selects the most significant EEG features, and using these selected features, a support vector machine (SVM) classifier performs the classification of MI signals. The proposed algorithm has been validated experimentally on two public BCI datasets (BCI Competition II Dataset III and BCI Competition IV Dataset 2b). The classification performance of the algorithm is quantified by the average accuracy and kappa coefficient, whose values are 80.7% and 0.615 respectively. The performance of the proposed algorithm is compared with standard feature selection methods based on Genetic Algorithm (GA), Principal Component Analysis (PCA), and ReliefF and performs better than these methods. Further, the proposed algorithm selects the lowest number of features and results in increased computational efficiency, which makes it a promising feature selection tool for an MI-based BCI system.

Segmentation and classification of medical images using texture-primitive features: Application of BAM-type artificial neural network.

The objective of developing this software is to achieve auto-segmentation and tissue characterization. Therefore, the present algorithm has been designed and developed for analysis of medical images based on hybridization of syntactic and statistical approaches, using artificial neural network (ANN). This algorithm performs segmentation and classification as is done in human vision system, which recognizes objects; perceives depth; identifies different textures, curved surfaces, or a surface inclination by texture information and brightness. The analysis of medical image is directly based on four steps: 1) image filtering, 2) segmentation, 3) feature extraction, and 4) analysis of extracted features by pattern recognition system or classifier. In this paper, an attempt has been made to present an approach for soft tissue characterization utilizing texture-primitive features with ANN as segmentation and classifier tool. The present approach directly combines second, third, and fourth steps into one algorithm. This is a semisupervised approach in which supervision is involved only at the level of defining texture-primitive cell; afterwards, algorithm itself scans the whole image and performs the segmentation and classification in unsupervised mode. The algorithm was first tested on Markov textures, and the success rate achieved in classification was 100%; further, the algorithm was able to give results on the test images impregnated with distorted Markov texture cell. In addition to this, the output also indicated the level of distortion in distorted Markov texture cell as compared to standard Markov texture cell. Finally, algorithm was applied to selected medical images for segmentation and classification. Results were in agreement with those with manual segmentation and were clinically correlated.