Advances in motor neuroprosthetics improve mobility in tetraplegics
Spinal cord injury, trauma of difficult solution
In Europe 330,000 people live with spinal cord lesions, and this number increases by 11,000 new cases per year. The main causes are work and motor vehicle accidents, especially among the youth, although in recent years the percentage of older adults has increased due to tumors and other causes. More than half the people with spinal cord injury are tetraplegic. The bilateral loss of the ability to grasp objects associated with this paralysis severely limits the capability of these individuals to live autonomously. It also limits their inclusion in society, as it generates a very strong dependence on caregivers and relatives. This is why one of the main priorities of these patients is to recover the ability to grasp with their hands.
An accident can result in complete or incomplete lesions. In both cases and focusing on the arms, if the degree of the lesion preserves the nervous system to the point of conserving sufficient voluntary control of the distal muscles to the elbow, surgical procedures can transfer muscles and tendons to recover a significant grasping function. However, if there are no residual motor functions or if the person refuses to undergo surgery (which includes a long period of post-surgery rehabilitation), current solutions are very limited and practically nonexistent for the recovery of mobility.
Pioneering neurotechnology with no surgical interventions
The ambition of the MoreGrasp project is to develop an upper limb neuroprosthesis to allow people with high-level SCI to carry out daily life activities autonomously. The three key technological-scientific principles that guide the project are:
Real life use: The motor neuroprosthetic device should be utilized continuously at the user’s home and during daily life activities, and therefore technologies from research laboratory settings are unsuitable.
No surgical procedures: Elimination of all invasive procedures, which means that all technology is applied without the necessity of complex interventions.
Individualization of prosthetics: The neuroprosthesis has to be customized at all levels to adjust to the different needs of each user.
This neuroprosthesis combines multiple technologies to enable the mobility of hands: functional electrical stimulation (FES) to activate the hand muscles; non-invasive brain-computer interface to decode intention and the type of movement/grasping in a natural manner; IoT technologies to provide information on the surroundings (for example, through instrumented objects); and artificial intelligence algorithms to combine all information and achieve shared control of the arm’s movement, with efficiency and robustness.
The basic functioning of the technology is described next: firstly, the user brings his/her hand close to the object he/she wishes to grasp (for example, a glass), and in a natural manner, thinks on how to grasp it. The brain-computer interface registers the signals from the brain’s motor cortex through an innovative EEG headset with dry sensors customized to the end user. Then, advanced EEG signal processing techniques and biomedical data analysis are used to obtain the type of movement intended. At the same time, information on the surroundings is registered. On the basis of all this information, the artificial intelligence engine provides instructions to the functional electrical stimulation system and personalized electrical currents are applied to the nerves of the user’s forearm. These currents activate the muscles that control the hand to create the intended grasp. The user can utilize the object and let it go by following the same process, and receives continuous sensory feedback by different means to compensate for the lack of sensibility (sense of touch).
Research, innovation and validation of neuroprosthetic limbs in real settings
The MoreGrasp project approaches significant scientific challenges through a combination of research, innovation and technology transfer between six universities, research centers and companies.
Firstly, the neuroprosthesis encompasses simultaneously four different scientific areas: robust and customized functional electrical stimulation, practical and customized neurotechnology and electroencephalogram (EEG) systems, new paradigms of brain-computer interfaces for the natural decoding of movements, and artificial intelligence systems and brain-machine interactions that integrate biosignals and context.
Secondly, it is not easy to accomplish a system that works without the help of experts with this degree of complexity and adaptation to each final user. This requires bringing the integration of all aforementioned technologies to an extremely high level of maturity to enable its simple and reliable use on a daily basis at home. Indeed, the final neuroprosthetics system is operated just with the assistance of relatives or usual caregivers through simple and intuitive interfaces on a tablet.
- Lastly, the MoreGrasp project aims at making a difference in the quality of life of tetraplegics. The final phase of the project, currently being carried out, consists of clinical trials that involve a validation of the neuroprosthetics with end users (those interested can enrol here). The process of the clinical trial is as follows: the users that fulfil the medical inclusion criteria (type of lesion, residual motor capabilities, no other medical conditions present such as stroke, epilepsy…) are examined by the research team to evaluate suitability. When selected, participants complete a 4- to 8-week training phase to strengthen arm muscle activity and to get familiarized with the brain-computer interface. At the same time, the FES prosthesis is customized along with the EEG system, and the biomedical signal treatment algorithms are tuned through automatic learning techniques. Once training is completed, the user receives his/her customized neuroprosthesis at home, which will be utilized daily for a period of 8 weeks. Although MoreGrasp personnel is always available and monitors the evolution, the users utilize the system autonomously, only with the help of their usual caregivers.
Reaching such a significant advance in non-invasive motor neuroprosthesis required the MoreGrasp research team to attain important achievements in different areas such as:
New practical EEG technologies. The first was a versatile EEG device based on sensors that worked with tap water for preliminary tests and training. The second was a minimalist EEG headset with dry sensors, optimized for daily use (very comfortable and easy to place), customizable and designed to measure sensory-motor states.
A new system of functional electrical stimulation based on an electrode network to selectively activate in real time specific muscle groups with compensation of arm rotation.
A new paradigm of brain-computer interface to achieve grasping, based on the intention of executing the movement in a natural manner and with all the associated EEG decoding algorithms.
Integration of the entire system in a multimodal professional platform for the development of brain-computer interfaces controlled by a tablet.
Next are the descriptions of those advances that are expected to have cross-sectional impact in different fields of research.
A pioneering EEG technology that is comfortable to wear during everyday life
MoreGrasp has developed two innovative EEG headsets. During the user-assessment and training phases, versatile EEG caps with water based electrodes are employed, with 16 and 32 channels, respectively. These EEG systems themselves suppose a step further in the use outside laboratory settings: they maintain the reliability and robustness of medical equipment but do not require electrolytic gels. In this way, users do not have to wash their hair after each use, which favors the acceptance of the technology. However, its shower cap-like design has usability and aesthetics barriers for everyday use. These caps require preparation periods that last 5-10 minutes, can result uncomfortable after long period of time, and their aesthetics usually generate rejection when use is not occasional (research, clinical evaluation or training phase).
For this reason, MoreGrasp in collaboration with multinational Nissan has developed an innovative minimalist EEG technology with dry sensors, optimized for the measurement of sensory and motor states. This is a very comfortable technology, that can be placed on average under two minutes, does not require the application of conductive substances and can be continuously utilized for up to 8 hours. The design is technological and more discreet than any other existing EEG technology. Regarding reliability, during the project it has been demonstrated that the minimalist EEG measures brain signals classically utilized in brain-computer interfaces such as MRCPs and ERD/ERS, with quality comparable to medical EEG systems. This neurotechnology developed by Bitbrain is a clear advance towards new EEG systems and brain-computer interfaces that are usable outside laboratory settings.
A brain-computer interface to decode different types of grasps
The project has developed an innovative neural interface based on a non-invasive brain machine interface that decodes the natural thoughts of SCI users and executes reaching and grasping tasks. The main difference to existing interfaces is that it utilizes information produced naturally when a person executes grasping movements, or in the case of people with SCI, when they think of executing. This is an important advance with respect to previous interfaces that require unnatural and repetitive thoughts to indicate actions (for example, thinking of moving the right foot repetitively to indicate the action of closing the hand).
The brain activity explored by the MoreGrasp project is produced both before (movement preparation activity) and during movement (activity related to the movement). When a person with SCI tries to execute movement, preparatory activity constituted by low frequency potentials (MRCPs) and motor desynchronizations (ERD/ERS) is produced mainly on the motor cortex. This preparatory activity precedes the movement itself, whose command is triggered by the brain, but does not reach the limb due to the spinal cord injury. Hence, the desired movement is not produced. However, this brain activity reflects the natural movement intention, presenting a different typology for each type of movement or grasping, and can be measured by EEG electrical signals and decoded by a brain-computer interface (read scientific publication).
Several problems had to be solved before actually changing the paradigm for people with spinal cord injury. Pattern recognition techniques were developed, capable of differentiating movement intention and the type of movement simultaneously, with high precision and in a completely continuous and natural way for the end user (scientific publication). Besides, electrical stimulation and the movement of the limb produce mechanical and electrical artifacts that contaminate the EEG, and must be filtered and corrected to increase the precision of the technology. A person with spinal cord injury progressively loses the motor brain patterns because of degeneration of the motor cortex due to non-use (scientific publication). This hinders the detection of brain patterns and requires special training and calibration procedures for customization of the procedure.
Functional electrical stimulation and AI
Functional electrical stimulation utilizes electrode arrays to produce electrical currents to stimulate the intact nerves of the arm and produce muscle contractions to move the paralyzed limb (turning the user’s own arm into a prosthetic arm). One of the challenges associated with this type of systems is achieving consistent and robust grasping action, independently of the position of the arm. In other words, traditional systems can achieve good patterns for the hand (for example, extending the thumb) with the arm in resting position. However, when the arm is rotated, the slight movements of the stimulation sensors and the nerves modify the stimulation pattern and destroy the grasping action. MoreGrasp addressed this issue by equipping the neuroprosthesis with a set of inertial units and a network of stimulation electrodes. In this way, rotation of the arm can be estimated through the inertial units and the stimulation patterns can be dynamically changed to compensate for such rotation. In the previous example, this means maintaining the extension of the thumb while the end user rotates his/her arm.
Also, the system is complemented by other sensors to infer context and provide feedback to the user. For example, everyday objects can be instrumented so that the technology understands which object is being interacted with, and adjusts its operation. This increases reliability and robustness of the system, also increasing the degree of acceptance by end-users. In this sense, the system is prepared to include any additional information required by the user such as a shoulder joystick (habitually utilized by these users) or more complex systems based on computer vision, projections and augmented reality. In the center of all this information is a system that, through artificial intelligence techniques, merges information to make the best decision at every moment and keeps the user informed on the state of the system through a smartwatch.
Conclusions and MoreGrasp consortium
The international MoreGrasp team is working with the objective of developing a motor neuroprosthetic limb that helps people with spinal cord injuries to carry out daily hand movements, at home and without expert supervision. Currently the neuroprosthesis is in clinical trial stage.
The MoreGrasp project has been coordinated with the Graz University of Technology, which has wide experience in biomedical engineering and in the design of brain-computer interfaces. Also at academic level, the University of Heidelberg has participated due to its expertise in spinal cord injury patients and functional electrical stimulation, and the University of Glasgow contributed with its knowledge on brain-machine interaction and artificial intelligence. At industrial level, Bitbrain led the development of new comfortable and reliable EEG technologies and the multimodal brain-computer interface platform due to its broad experience in neural engineering and neurotechnology for brain controlled robots. Medel was responsible for the development of the functional electrical stimulation, and KnowCenter was responsible for the web platform for enrollment of patients.
The Bitbrain team