Utilization of Electromyographic Signal Classification for Predicting Hand Gestures and Grasping Forces Through Artificial Neural Networks

Derya Karabulut; Suzan Cansel Doğru

Makale detayları

Utilization of Electromyographic Signal Classification for Predicting Hand Gestures and Grasping Forces Through Artificial Neural Networks

Yazarlar:

Derya Karabulut | Suzan Cansel Doğru

Yayıncı:

Gazi University

Ana Başlıklar:

Biomechanic

Dergi:

Gazi University Journal of Science Part A: Engineering and Innovation Vol.12 Issue 2, pp.652-664

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001348253

pISSN:

2147-9542

Anahtar kelimeler:

Özet:

 Electromyographic EMG signals have gained significant recognition for their potential applications in various fields, particularly in the classification of hand gestures and grasping force prediction. These signals provide critical information that can be utilized to develop systems capable of interpreting human intentions, making them indispensable in areas such as assistive technology and human-computer interaction. The primary objective of this study was to assess the effectiveness of different time-domain features extracted from EMG signals in predicting hand gestures and grasping force simultaneously. The study specifically tested features such as Root Mean Square RMS, Variance of EMG VAR, Waveform Length WL, Integrated EMG IEMG, Difference Absolute Standard Deviation Value DASDV, and Difference Absolute Mean Value DAMV, employing an artificial neural network ANN to evaluate their ability to predict hand movements and grasping force. EMG data were collected from the flexor carpi radialis muscle, which plays a significant role in hand movement control. After extracting the relevant time-domain features from the EMG signals, these were input into an ANN for further analysis. The results demonstrated that the RMS feature provided the highest accuracy for both prediction of hand gesture success rate 90.0 for resting position, 93.3 for wrist flexion, and 86.7 for hand pronation and grasping force 0.09 RMSD and 0.89 PCC values for resting position, 0.15 RMSD and 0.85 PCC for wrist position. These findings underscore the potential application of these EMG-derived features in practical systems, particularly in the development of myoelectric-controlled prostheses. Such advancements hold promise for enhancing the functionality and intuitiveness of prosthetic devices, offering users more efficient and effective control, and thus improving their overall quality of life in assistive technology contexts.

 Electromyographic EMG signals have gained significant recognition for their potential applications in various fields, particularly in the classification of hand gestures and grasping force prediction. These signals provide critical information that can be utilized to develop systems capable of interpreting human intentions, making them indispensable in areas such as assistive technology and human-computer interaction. The primary objective of this study was to assess the effectiveness of different time-domain features extracted from EMG signals in predicting hand gestures and grasping force simultaneously. The study specifically tested features such as Root Mean Square RMS, Variance of EMG VAR, Waveform Length WL, Integrated EMG IEMG, Difference Absolute Standard Deviation Value DASDV, and Difference Absolute Mean Value DAMV, employing an artificial neural network ANN to evaluate their ability to predict hand movements and grasping force. EMG data were collected from the flexor carpi radialis muscle, which plays a significant role in hand movement control. After extracting the relevant time-domain features from the EMG signals, these were input into an ANN for further analysis. The results demonstrated that the RMS feature provided the highest accuracy for both prediction of hand gesture success rate 90.0 for resting position, 93.3 for wrist flexion, and 86.7 for hand pronation and grasping force 0.09 RMSD and 0.89 PCC values for resting position, 0.15 RMSD and 0.85 PCC for wrist position. These findings underscore the potential application of these EMG-derived features in practical systems, particularly in the development of myoelectric-controlled prostheses. Such advancements hold promise for enhancing the functionality and intuitiveness of prosthetic devices, offering users more efficient and effective control, and thus improving their overall quality of life in assistive technology contexts.

 Electromyographic EMG signals have gained significant recognition for their potential applications in various fields, particularly in the classification of hand gestures and grasping force prediction. These signals provide critical information that can be utilized to develop systems capable of interpreting human intentions, making them indispensable in areas such as assistive technology and human-computer interaction. The primary objective of this study was to assess the effectiveness of different time-domain features extracted from EMG signals in predicting hand gestures and grasping force simultaneously. The study specifically tested features such as Root Mean Square RMS, Variance of EMG VAR, Waveform Length WL, Integrated EMG IEMG, Difference Absolute Standard Deviation Value DASDV, and Difference Absolute Mean Value DAMV, employing an artificial neural network ANN to evaluate their ability to predict hand movements and grasping force. EMG data were collected from the flexor carpi radialis muscle, which plays a significant role in hand movement control. After extracting the relevant time-domain features from the EMG signals, these were input into an ANN for further analysis. The results demonstrated that the RMS feature provided the highest accuracy for both prediction of hand gesture success rate 90.0 for resting position, 93.3 for wrist flexion, and 86.7 for hand pronation and grasping force 0.09 RMSD and 0.89 PCC values for resting position, 0.15 RMSD and 0.85 PCC for wrist position. These findings underscore the potential application of these EMG-derived features in practical systems, particularly in the development of myoelectric-controlled prostheses. Such advancements hold promise for enhancing the functionality and intuitiveness of prosthetic devices, offering users more efficient and effective control, and thus improving their overall quality of life in assistive technology contexts.

 Electromyographic EMG signals have gained significant recognition for their potential applications in various fields, particularly in the classification of hand gestures and grasping force prediction. These signals provide critical information that can be utilized to develop systems capable of interpreting human intentions, making them indispensable in areas such as assistive technology and human-computer interaction. The primary objective of this study was to assess the effectiveness of different time-domain features extracted from EMG signals in predicting hand gestures and grasping force simultaneously. The study specifically tested features such as Root Mean Square RMS, Variance of EMG VAR, Waveform Length WL, Integrated EMG IEMG, Difference Absolute Standard Deviation Value DASDV, and Difference Absolute Mean Value DAMV, employing an artificial neural network ANN to evaluate their ability to predict hand movements and grasping force. EMG data were collected from the flexor carpi radialis muscle, which plays a significant role in hand movement control. After extracting the relevant time-domain features from the EMG signals, these were input into an ANN for further analysis. The results demonstrated that the RMS feature provided the highest accuracy for both prediction of hand gesture success rate 90.0 for resting position, 93.3 for wrist flexion, and 86.7 for hand pronation and grasping force 0.09 RMSD and 0.89 PCC values for resting position, 0.15 RMSD and 0.85 PCC for wrist position. These findings underscore the potential application of these EMG-derived features in practical systems, particularly in the development of myoelectric-controlled prostheses. Such advancements hold promise for enhancing the functionality and intuitiveness of prosthetic devices, offering users more efficient and effective control, and thus improving their overall quality of life in assistive technology contexts.

Makale detayları

Utilization of Electromyographic Signal Classification for Predicting Hand Gestures and Grasping Forces Through Artificial Neural Networks

Yazarlar:

Derya Karabulut | Suzan Cansel Doğru

Yayıncı:

Gazi University

Ana Başlıklar:

Biomechanic

Dergi:

Gazi University Journal of Science Part A: Engineering and Innovation Vol.12 Issue 2, pp.652-664

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001348253

pISSN:

2147-9542

Electromyography

Artificial Neural Network

Root Mean Square

Classification

Sayı

Makale detayları

Utilization of Electromyographic Signal Classification for Predicting Hand Gestures and Grasping Forces Through Artificial Neural Networks

Yazarlar: Derya Karabulut | Suzan Cansel Doğru

Yayıncı: Gazi University

Ana Başlıklar: Biomechanic

Dergi: Gazi University Journal of Science Part A: Engineering and Innovation Vol.12 Issue 2, pp.652-664

Belge Tipi: Makale

Yayın Tarihi: 6/2025

Belge Dili: İngilizce

Document ID (Org No.): 2001348253

pISSN: 2147-9542

Electromyography

Artificial Neural Network

Root Mean Square

Classification

Sayı

Yazarlar:

Derya Karabulut | Suzan Cansel Doğru

Yayıncı:

Gazi University

Ana Başlıklar:

Biomechanic

Dergi:

Gazi University Journal of Science Part A: Engineering and Innovation Vol.12 Issue 2, pp.652-664

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001348253

pISSN:

2147-9542