interview

Professor Sofiane Achiche of Ecole Polytechnique de Montreal, Canada, talks to us about the work behind the paper ‘Detecting muscle contractions using strain gauges’, page 1836 Professor Sofiane Achiche Myoelectric prostheses are artificial limbs, controlled by surface electromyography signals, which are produced by the residual muscles of amputees. They have the potential to offer intuitive control, and could allow multiple degrees of freedom. The current challenge to control phantom limbs with myoelectric prostheses is detecting information about muscle activation in the residual limbs. Nowadays, the use of electromyography measurements of muscles is increasingly used for the monitoring of upper limb motor function. However, the quality of electromyography for muscle motor functional monitoring depends on its reliability. However, electromyography still has some drawbacks. Amputees who do not show much electrical activity in the remaining muscles, a phenomenon often noticeable in patient whose amputation dates, present a difficult obstacle. A challenging option is to replace the electrical signals with skin deformations around the muscle's, so the muscles activation will be detected by analysing these deformations. The aim of our paper was to propose a novel device capable of detecting muscle contractions, with the incentive to control myoelectric prostheses. The key strategy was detection of the the small skin deformations by strain gauge, and predicting the intents of the main upper limb movements during the contractions of these different muscles. However, this project was borne of the fact that we managed to detect ghost movement intentions in amputees, such as the opening of the hand, for which they no longer have the muscles for. This was achieved with sensors (initially electromyography) placed on the residual upper limb. However, when we asked the subjects to imagine opening/closing their phantom hand, the muscles in their residual limb moved too little. So we wanted to check if it was possible to measure the deformation of the skin - or even the sliding of the skin - on the strain gauge rather than the EMG. And it worked! This study confirms that strain gauges are a suitable sensor alternative for detecting muscle activation, characterised by their simplicity and low cost. But most of all, the greatest interest seems to be in obtaining a signal equivalent to the filtered EMG, with just a sampling frequency of 5 Hz, instead of 1000 Hz. In a context where myoelectric prostheses have more and more sensors, especially for novel prostheses using classification and machine learning to detect the movement intents in phantom limbs, strain gauges could be very interesting, particularly as the prostheses processors are now struggling to manage real-time signal classification. With strain gauges working with sampling frequencies around 200 times lower, the number of sensors in prostheses could potentially increase by a factor 200. This could lead to refined prostheses, which today are still considered futuristic! This would considerably boost the fields of research into prostheses development, the control of robotic arms, and methods of classification and machine learning for these applications. For my colleague, Prof. Maxime Raison, and I, the new opportunities are altruistic in nature. It will be a pleasure to work, in the next months and years, towards helping amputees and the large variety of people in rehabilitation, who suffer with conditions such as cerebral palsy, spinal cord injuries, muscular dystrophy, delivering the message that this technology is moving and developing rapidly. In association with the Laboratory of Mechatronics Design, and the Rehabilitation Engineering Chair Applies to Pediatrics of École Polytechnique de Montréal, we developing “smart” myoelectric prostheses based on novel sensors and machine learning. Building on this expertise, we are also designing an exoskeleton dedicated to helping children in rehabilitation, and on the control of assistive robotic arms using vision, and even thought.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.