Design of a BMI for tetraplegic patients

Paralysis can completely change a person's life. Depending on the severity, patients suffering from spinal cord injury may loss their ability to move or speak. But technology can help to restore their autonomy. In this master thesis, brain-machine interfaces (BMIs) are explored. A BMI bypasses the broken communication link between the brain and the muscles by measuring and decoding the brain activity to either stimulate the limbs if the patient's body is not damaged, or control an external device, like an effector or a computer. This work continues 3 years of previous development in the design of an EEG-based BMI. The first year, a prototype was build to demonstrate the feasibility from acquisition to real-time classification of user's movements (left versus right arm). Their system was composed of a PCB with a FPGA as main component, and a 3D-printed helmet with dry electrodes. The two next years, they investigated processing techniques and combination of algorithms for the different stages of the BMI in order to improve the classification accuracy. This year, the application side was considered. In the idea of reaching wheelchair control, a video game has been developed as proof of concept. The number of classified movements was increased to 3 (left hand to turn left, right hand to turn right, both feet to move forward). An asynchronous BMI with sufficient response time was targeted, and a cumulative incremental control strategy was proposed to smooth the vehicle displacements, based on the discrete commands outputted by the BMI.