Relevance of Virtual Reality Assessment of Control Interfaces for a Neuroprosthesis That Stimulates the Upper Limbs of Quadriplegic Individuals
Relevance of Virtual Reality Assessment of Control Interfaces for a Neuroprosthesis That Stimulates the Upper Limbs of Quadriplegic Individuals
With a view to implanting a neuroprosthesis for the upper limb in individuals with complete tetraplegia, this study aims to validate virtual reality as a simulation tool for evaluating and optimizing the piloting interfaces of these type of assistive devices. The operating principle of the device we are interested in is to provide a set of predefined functional electrical stimulation (FES) configurations that activate specific hand movements (hand opening, palmar grasp, key grip, etc.), from which the user selects the one that is suitable for performing a task. The user control is usually based on commands from the contralateral limb (pressing a button, shoulder movement, or voluntary muscle contractions). These stereotypical and unintuitive commands hinder any possibility of bimanual tasks. Hands-free voice interfaces have been tested but have contextual limitations, particularly in terms of discretion or usability for certain activities such as eating. Furthermore, it is difficult to evaluate the performance of control interfaces and adjust them prior to the implantation of the stimulation neuroprosthesis. The aim of the I-GRIP project is to establish a methodology that is sufficiently realistic to enable people to envision their future use of a neuroprosthesis. Such a tool would also enable future candidates for implantation to better understand the device's potential. This approach would also make it possible to customize the technology prior to implantation (choice and adjustment of control interfaces, training, configuration of algorithms for analyzing movements evoked by stimulation, etc.).
Our main hypothesis is that two control interfaces (HMI1 and HMI2) will allow the user to control the completion of a grasping task (approach, grasp, hold) for each target object in the virtual environment simulating electrical stimulation of the forearm muscles.
Immersive, semi-immersive, or non-immersive VR has mainly been used for therapeutic rehabilitation purposes in patients with spinal cord injuries who have motor impairments in their upper limbs. A recent literature review in 2024 highlights a strong trend emerging from the analysis of all publications on the subject, namely an analytical gain in strength with no impact on manual dexterity, this gain being all the more significant the longer the training lasted (> 15 hours in total). In the field of functional electrical muscle stimulation, which is of interest to us here, Ajiboye et al. 2017 proposed post-implantation training of an upper limb (UL) stimulation neuroprosthesis in a quadriplegic patient under the control of a Brain Computer Interface (BCI) and compared object grasping performance in immersive VR and non-VR. Given that VR training was longer than non-VR training, movements were found to be smoother and more precise in VR. The uniqueness of this experiment lies in having established proof of concept around a subject's ability to control single and multi-coordinated movements with 80 to 100% accuracy using their thoughts, first using a virtual upper limb (UL) and then their own UL activated by FES to drink a cup of coffee and bring food to their mouth. It also highlighted the real value of immersive virtual reality in learning movements.
A recent publication of a clinical case study focused on the use of a Leap Motion optical hand tracking controller in a semi-immersive environment [Ahmed Al Nattah 2024]. Leap Motion, which is mainly used in kinematic assessments, is introduced in this publication for the purpose of rehabilitating the hand of an incomplete quadriplegic participant equipped with sensors and invited to perform exercises of increasing difficulty in custom-built video games offering both visual and auditory feedback. Beyond the promising therapeutic benefits observed in this single case study, this semi-immersive experience can be transposed into full immersion in applications such as the optimization and customization of neuroprosthesis control interfaces. VR is playing an increasingly decisive role in movement rehabilitation. It is part of the therapeutic arsenal that now aims to offer a fun, meaningful, and motivating approach to encourage the repetition of gestures and movements and, in fact, learning. However, its use as a tool for designing and evaluating the performance of assistive technology control interfaces has been little explored in studies, which justifies the approach adopted in this project.
With a view to implanting a neuroprosthesis for the upper limb in individuals with complete tetraplegia, this study aims to validate virtual reality as a simulation tool for evaluating and optimizing the piloting interfaces of these type of assistive devices. Such an environment would ultimately enable future users to prepare for the use of these devices and to project themselves into their use. This approach would also make it possible to customize the technology prior to implantation (choice and adjustment of control interfaces, training, configuration of algorithms for analyzing movements evoked by stimulation, etc.). In general, it is important to advance control interfaces to improve their ergonomics and usability. Here, we wish to show that virtual reality allows participants to realistically project themselves into an experience of using a neuroprosthesis that controls wrist extension and finger flexion/extension movements.
Participants will be equipped with a virtual reality headset that will project a virtual scene with various everyday objects and a representation of their hand. They will have to move their arm towards one of the objects in the scene, then trigger hand movements using one of the available control interfaces to grasp the object and manipulate it. Two methods of controlling the opening and closing movements of the fingers will be tested: HMI1: two physical buttons activated by pressing with the opposite hand to send commands, and HMI2 - shared control: the reaching movement of the arm towards the object is tracked using a video camera and automatically select the appropriate opening and gripping movements. HMI1 therefore requires a stereotypical action (pressing a button), while HMI2 allows automatic selection of grasping movements by observing the natural movements of approaching the object, which are observed by a camera and interpreted by an algorithm.
The experimental aspects of the study will be conducted over 2 sessions:
Inclusion Criteria:
Exclusion Criteria:
• Participant deprived of liberty (by judicial or administrative decision).