Hybrid Deep Learning Approaches for sEMG Signal-Based Lower Limb Activity Recognition

Vijayvargiya, Ankit; Singh, Bharat; Kumar, Rajesh; Desai, Usha; Hemanth, D. Jude

doi:10.1155/2022/3321810

Cited by 13 publications

(6 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In 2022, Vijayvargiya et al [25] developed hybrid deep learning models for this purpose, incorporating discrete wavelet transform for noise suppression and employing convolutional neural networks of sequential learning, alongside long short-term memory or gated recurrent units of ordered learning.…”

Section: Literature Reviewmentioning

confidence: 99%

See 1 more Smart Citation

Integrated Sensor Fusion and Multi-Modal Hierarchical Neural Network for Activity Recognition in Lower Limb Prosthetics

S. Prakash

2024

jes

View full text Add to dashboard Cite

Activity recognition plays pivotal role in enhancing functionality for prosthetic devices, ensuring seamless integration with users' movements. However, the complexity arises from the diverse data sources, including acceleration, angular velocity, joint angles, orientation, electromyography (EMG), and marker data, necessitating a robust approach to overcome challenges in information integration. The primary challenge lies in the effective utilization of multiple sensor modalities, each with unique characteristics and potential noise sources. The proposed solution addresses this by employing advanced sensor fusion techniques, such as Kalman filtering, during data collection. Synchronization and resampling ensure temporal consistency, while noise reduction techniques, such as low-pass filters, mitigate signal distortions. To further refine the process, a hybrid optimization-based feature selection Adaptive Step Size in Marine Predators Algorithm (ASSMPA) is introduced, focusing on marker data features. ASSMPA synergizes Marine Predators Algorithm (MPA) and Pathfinder Algorithm (PFA) for optimal feature selection in marine predator pathfinding tasks. The Feature Fusion step integrates attention mechanisms to dynamically weigh the significance of different sensor modalities during the fusion process. This strategic fusion enhances the overall performance of the Multi-Modal Hierarchical Neural Network (MMHNN). The proposed model is implemented using Python.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

“…The feature set {𝑓𝑓 1 ⊕ 𝑓𝑓 2 ⊕ … ⊕ 𝑓𝑓 𝑛 } in Eq. [25] encompasses various combinations, incorporating deep learning, color, shape, and texture-based features. Features are balanced by assigning weights based on their significance, using Eq.…”

Section: Feature Fusionmentioning

confidence: 99%

Integrated Sensor Fusion and Multi-Modal Hierarchical Neural Network for Activity Recognition in Lower Limb Prosthetics

S. Prakash

2024

jes

View full text Add to dashboard Cite

show abstract

“…Anomaly detection methods have removed abnormal data with the light gradient boosting machine to reach 98% accuracy. Furthermore, it has been shown that EMG signals can predict lower limb activities for normal and OA patients using a hybrid deep learning model [34]. Interestingly [32][33][34], studies have used the same data sets, which comprise EMG signals for 11 normal subjects and 11 subjects with knee OA.…”

Section: Introductionmentioning

confidence: 99%

“…Furthermore, it has been shown that EMG signals can predict lower limb activities for normal and OA patients using a hybrid deep learning model [34]. Interestingly [32][33][34], studies have used the same data sets, which comprise EMG signals for 11 normal subjects and 11 subjects with knee OA. The data have been taken from the biceps femoris (BF), vastus medialis (VM), rectus femoris (RF), and semitendinosus (ST) muscles while the subject is walking, sitting, and standing.…”

Section: Introductionmentioning

confidence: 99%

Combining enhanced spectral resolution of EMG and a deep learning approach for knee pathology diagnosis

Khader,

Zyout,

Al Fahoum

2024

PLoS ONE

View full text Add to dashboard Cite

Knee osteoarthritis (OA) is a prevalent, debilitating joint condition primarily affecting the elderly. This investigation aims to develop an electromyography (EMG)-based method for diagnosing knee pathologies. EMG signals of the muscles surrounding the knee joint were examined and recorded. The principal components of the proposed method were preprocessing, high-order spectral analysis (HOSA), and diagnosis/recognition through deep learning. EMG signals from individuals with normal and OA knees while walking were extracted from a publicly available database. This examination focused on the quadriceps femoris, the medial gastrocnemius, the rectus femoris, the semitendinosus, and the vastus medialis. Filtration and rectification were utilized beforehand to eradicate noise and smooth EMG signals. Signals’ higher-order spectra were analyzed with HOSA to obtain information about nonlinear interactions and phase coupling. Initially, the bicoherence representation of EMG signals was devised. The resulting images were fed into a deep-learning system for identification and analysis. A deep learning algorithm using adapted ResNet101 CNN model examined the images to determine whether the EMG signals were conventional or indicative of knee osteoarthritis. The validated test results demonstrated high accuracy and robust metrics, indicating that the proposed method is effective. The medial gastrocnemius (MG) muscle was able to distinguish Knee osteoarthritis (KOA) patients from normal with 96.3±1.7% accuracy and 0.994±0.008 AUC. MG has the highest prediction accuracy of KOA and can be used as the muscle of interest in future analysis. Despite the proposed method’s superiority, some limitations still require special consideration and will be addressed in future research.

show abstract

“…The approach was evaluated on four lower limb movements from twenty people and exhibited 95.82% and 97.44% accuracy for two muscles. Vishu Gupta et al [38] projected that neuromuscular problem diagnosis and kinesiological research require surface EMG (sEMG) signals employing hybrid wavelet-ensemble empirical mode decomposition (WD-EEMD). Lower extremity exoskeleton robots can be more lifelike using a wavelet packet transformbased sliding window difference average filtering feature extract algorithm and UKFNN identification algorithm.…”

Section: Introductionmentioning

confidence: 99%

Variational Mode Decomposition and Empirical Wavelet Transform-Based Feature Extraction and Ensemble Classifier for Lower Limb Movement Prediction With Surface Electromyography Signal

Rani,

Hashmi,

Muhammad

2024

IEEE Access

View full text Add to dashboard Cite

Surface Electromyography (sEMG) signal classification and analysis have attracted particular attention because of their many biological applications. These can be utilized for kinesiological research, motion intention recognition, human-machine interaction, rehabilitation, and disease diagnosis. Due to the interference of noise, these signals are challenging to process. To attenuate noise interference from sEMG signals, a denoising method based on Variational Mode Decomposition (VMD) with selected Intrinsic Mode Functions (IMFs) based on Correlation Coefficient (CC) and Empirical Wavelet Transform (EWT) is proposed in this study. The suggested method applies the EWT to the specific IMFs obtained by VMD, decomposing the raw signal with noise into several IMFs based on correlation coefficients. To increase the signal separability, stability and to extract the low-frequency information of the signal, apply EWT on selected IMFs; it decomposes IMFs into the multiresolution analysis (MRA). Following segmentation, 16 Features are recovered using the time domain, frequency domain, and entropy with an overlapping window technique of 250ms and 50% overlap. Later, four machine learning classifiers-Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), Naive Bayes (NB), and Ensemble Subspace k-nearest Neighbor (KNN)are used to identify the three lower limb movements. The results obtained for an Ensemble Subspace KNN classifier have a 98.9% accuracy, a 98.8% F1-Score, and a 98.7% sensitivity. Lastly, machine learning (ML) is used in this work to create a new feature extraction technique for lower limb activity prediction based on VMD-EWT with CC.

show abstract

Hybrid Deep Learning Approaches for sEMG Signal-Based Lower Limb Activity Recognition

Cited by 13 publications

References 40 publications

Integrated Sensor Fusion and Multi-Modal Hierarchical Neural Network for Activity Recognition in Lower Limb Prosthetics

Integrated Sensor Fusion and Multi-Modal Hierarchical Neural Network for Activity Recognition in Lower Limb Prosthetics

Combining enhanced spectral resolution of EMG and a deep learning approach for knee pathology diagnosis

Variational Mode Decomposition and Empirical Wavelet Transform-Based Feature Extraction and Ensemble Classifier for Lower Limb Movement Prediction With Surface Electromyography Signal

Contact Info

Product

Resources

About