Cooperation optimizes control of prostheses

Cooperation optimizes control of prostheses

Interface improves capture and use of myoelectric signals in people who have lost a hand or part of a forearm

Interface improves capture and use of myoelectric signals in people who have lost a hand or part of a forearm

Have you, the reader who has come this far, ever stopped to think about the role of your hands in your reading experience? Perhaps you are holding the printed edition of the newspaper, your thumbs gently pressing the sides of these pages, while your index, middle, ring and little fingers bend to support the front and back covers. Or perhaps this is the online version of the article and, with a smartphone firmly resting in your palm, you slide your finger across the screen to “scroll” the website page. But it may also be that you do not have one or neither of your hands, and in that case, reading these words may not be so easy without some kind of assistance.

For most people, precise control of their own gestures is so instinctive that the effort required to bend a single finger becomes almost unconscious. However, for those who have suffered amputations, controlling a prosthetic hand is a significant challenge, requiring a period of adaptation and training, in addition to understanding that these devices may not achieve the desired level of flexibility. Now, however, a project by the Center for Biomedical Engineering (CEB) at Unicamp is close to making the control of these devices more precise and sophisticated, through an interface that improves the capture and use of myoelectric signals.

Skeletal muscle produces myoelectric signals when the subject exerts force to perform a movement. Initiated in the brain, this command propagates through neurons in the spinal cord or brain stem – the motor neurons – until it reaches the muscle fibers. These, in turn, act as an amplifier of the signal, which can be read by electrodes. Since different gestures send different patterns and amplitudes, the electrodes can identify the intention of each movement individually. “So the idea is that we can use the muscles that originally performed those movements to create the interface. They collect the signal, process it and make the intention of the movement act on the prosthesis”, explains professor Leonardo Abdala Elias, who coordinates the research.

Although the use of signals of this type is not new in the production of prosthetics, the Unicamp project uses a technique that is currently absent from commercial devices. Called high-density electromyography, this resource uses a much larger number of electrodes in a small region of the body, allowing the collection of a larger volume of signals relating to an area of ​​the muscle. In the case of the system under development, 128 channels map the entire recorded region, which means 128 signals captured to identify the performance of the gesture. By way of comparison, typical prosthetics use between two and eight channels to read the intention of a movement.

Funded by the Labor Prosecutor's Office (MPT), the technology, according to the researchers' goal, will serve people who have lost their hand or part of their forearm in work accidents. However, there is still a long way to go. At this point, the project already has a well-validated and robust acquisition, recording and signal processing system, but it is still necessary to improve the mechanical aspect of the prosthesis – produced using 3D printing techniques from a model open-hardware of bionic hand provided by the company Open Bionics. This includes both the improvement of the movement capacity and the development of accessories for attaching the prosthesis to the arm of the amputee user.

“At this stage, we are able to extract the intention of the movements with one degree of freedom. The human hand, however, has about 25 degrees of freedom, which refer to the ways in which it can move at each joint, activating the muscles in a synergistic manner,” says engineer Ricardo Molinari, who is working on his doctorate within the project. Responsible for understanding the electrophysiological characteristics of the neuromuscular system and developing the prosthesis’ movement system, Molinari explains that the group of scientists is currently working with a simplified version of the product. “The prototype can perform the main movements from a functional point of view, such as strength grip, precision and pinch with three fingers, but performing one type of movement at a time, which is a very preliminary stage,” he explains.

Next Steps

One of the main challenges in improving the flexibility of the prosthesis is to promote simultaneous and proportional control of movements, allowing the mechanical hand to move the fingers individually and in combination. In addition, there is a need to improve the algorithm responsible for identifying the intention of movements from the 128 channels, providing simultaneous control of a greater number of degrees of freedom. It is worth noting, however, that the researchers' goal is not limited to creating the most flexible mechanical prosthesis on the market – something already available, albeit at substantial prices – but to build an open acquisition and control system capable of controlling both simple and more sophisticated prostheses.

Scientists still need to test the technology on people who do not have hands, however, something that will be done starting next year. So far, tests with non-amputee volunteers have allowed them to train the algorithm that controls the interface and obtain normative and standardized data on the kinematics of the human hand. This survey has been conducted by postdoctoral researcher Guilherme Augusto Gomes. To collect these movements, doctoral student Valéria Carrillo created a virtual model for the laboratory and the hand prosthesis with controls similar to those used in Molinari's physical system.

In the experiments, eight infrared cameras positioned in the Neuromechanics and Neurorehabilitation Division of the CEB Neuroengineering Research Laboratory (NER) detect retroreflective markers attached to a user's body. In addition to reconstructing the biomechanics of the human body in this environment, the technology seeks to allow amputee volunteers to practice using the prosthesis in an immersive virtual environment – ​​using 3D glasses – before beginning the process with the physical prototype.

According to Elias, the training in the virtual environment aims to help volunteers who have not used their skeletal muscles for years, helping them adapt to the use of technology to control a new limb. “By doing this in an immersive environment rather than on a screen, we intend to provide several interaction alternatives and allow the patient to be more engaged in the training. We will use functional tests in a virtual environment to train the person and, in parallel, introduce the use of the physical prosthesis, until the time of the actual prosthesis fitting,” concludes the professor.