Decodificación de Movimientos Individuales de los Dedos y Agarre a Partir de Señales Mioeléctricas de Baja Densidad
Fecha
Autores
Villarejo Mayor, John J.
Mamede Costa, Regina
Frizera Neto, Anselmo
Freire Bastos, Teodiano
Directores
Unidades organizativas
Handle
https://riunet.upv.es/handle/10251/143518
Cita bibliográfica
Villarejo Mayor, JJ.; Mamede Costa, R.; Frizera Neto, A.; Freire Bastos, T. (2017). Decodificación de Movimientos Individuales de los Dedos y Agarre a Partir de Señales Mioeléctricas de Baja Densidad. Revista Iberoamericana de Automática e Informática industrial. 14(2):184-192. https://doi.org/10.1016/j.riai.2017.02.001
Titulación
Resumen
[ES] Uno de los principales retos en el diseño de prótesis de mano es poder establecer un control intuitivo que reduzca el esfuerzo del usuario durante su entrenamiento. Este trabajo presenta un esquema para identificar tareas de motricidad fina de la mano, agrupadas en movimientos de los dedos individuales y gestos para el agarre de objetos el cual se ha validado con sujetos amputados. Se han comparado diferentes métodos de selección de características y clasificadores para el reconocimiento de patrones mioeléctricos, utilizando cuatro electrodos superficiales. Las características de las señales en el dominio del tiempo y la frecuencia se han combinado con métodos no lineales basados en análisis de fractales, mostrando una diferencia significativa en comparación con los métodos expuestos en la literatura para clasificar tareas de fuerza. Los resultados con amputados mostraron una exactitud de hasta 99,4% en los movimientos individuales de los dedos, superior a la obtenida con los gestos de agarre, de hasta 93,3%. El sistema ha obtenido una tasa de acierto promedio de 86,3% utilizando máquinas de soporte vectorial (SVM), seguido muy de cerca por K-vecinos más cercanos (KNN) con 83,4%. Sin embargo, KNN ha obtenido un mejor rendimiento global, debido a que es más rápido que SVM, lo que representa una ventaja para aplicaciones en tiempo real. El método aquí propuesto ofrece una mayor funcionalidad en el control de prótesis de mano, lo que mejoraría su aceptación por parte de los amputados.
[EN] Intuitive prosthesis control is one of the most important challenges in order to reduce the user effort in learning to use an artificial hand. This work presents the development of a myoelectric pattern recognition system for myoelectric weak signals able to discriminate dexterous hand movements using a reduced number of electrodes. The system was evaluated in six forearm amputees and the results were compared with the performance of able-bodied subjects. Different methods were analyzed to classify individual fingers flexion, hand gestures and different grasps using four electrodes and considering the low level of muscle contraction in these tasks. Multiple features of sEMG signals were also analyzed considering traditional magnitude-based features and fractal analysis. Statistical significance was computed for all the methods using different set of features, for both groups of subjects (able-bodied and amputees). For amputees, results showed accuracy up to 99.4% for individual finger movements, higher than the achieved by grasp movements, up to 93.3%. Best performance was achieved using support vector machine (SVM), followed very closely by K-nearest neighbors (KNN). However, KNN produces a better global performance because it is faster than SVM, which implies an advantage for real-time applications. The results show that the method here proposed is suitable for accurately controlling dexterous prosthetic hands, providing more functionality and better acceptance for amputees.
[EN] Intuitive prosthesis control is one of the most important challenges in order to reduce the user effort in learning to use an artificial hand. This work presents the development of a myoelectric pattern recognition system for myoelectric weak signals able to discriminate dexterous hand movements using a reduced number of electrodes. The system was evaluated in six forearm amputees and the results were compared with the performance of able-bodied subjects. Different methods were analyzed to classify individual fingers flexion, hand gestures and different grasps using four electrodes and considering the low level of muscle contraction in these tasks. Multiple features of sEMG signals were also analyzed considering traditional magnitude-based features and fractal analysis. Statistical significance was computed for all the methods using different set of features, for both groups of subjects (able-bodied and amputees). For amputees, results showed accuracy up to 99.4% for individual finger movements, higher than the achieved by grasp movements, up to 93.3%. Best performance was achieved using support vector machine (SVM), followed very closely by K-nearest neighbors (KNN). However, KNN produces a better global performance because it is faster than SVM, which implies an advantage for real-time applications. The results show that the method here proposed is suitable for accurately controlling dexterous prosthetic hands, providing more functionality and better acceptance for amputees.
Palabras clave
Señales electromiográficas, Prótesis de miembro superior, Reconocimiento de patrones, Tareas de destreza de la mano, Myoelectric signals, Upper-limb prosthesis, Superficial electromyography low density, Dexterous hand gestures, Pattern recognition
ISSN
1697-7912
ISBN
Fuente
Revista Iberoamericana de Automática e Informática industrial
DOI
10.1016/j.riai.2017.02.001
Versión del editor
https://doi.org/10.1016/j.riai.2017.02.001
dc.description.uri
Código de Proyecto
Agradecimientos
Este trabajo ha sido patrocinado por CAPES y FAPES/Brasil
(Proyecto Número 007/2014: Use of Robotics and Assistive
Technology for Children and Adults with Disabilities).