Design of a BMI for tetraplegic patients

A Brain-Computer Interface (BCI) is a system that translates a user’s brain activity into messages or commands for an interactive application. Such a system offers the possibility to take control of a device by thought. Therefore this type of system is totally adapted to help people with physical disabilities. This is the case of tetraplegic patients, who have lost the ability to move their limbs. BCIs represent a relatively recent technology that is experiencing a rapid growth due to recent advances in electronics and signal processing. In this context, this Master Thesis aims at exploring the field of BCI and develop a system that would be able to detect imaginary movements. This work continues the those done previously by implementing a complete BCI application in real-time on a FPGA, able to collect data and tell if a movement of the right arm, left arm or not movement at all has been done. To achieve this goal, different steps have been completed, each one corresponding to a chapter of this work. The field of BCI is a complex area of study and we needed to be up to date with the trends and applications up to now. This work thus starts with a state of the art and review of the anatomy of the brain and different brain signals. It also continues with the exploration of the different existing algorithms and techniques for the implementation of such a system. An experimental board received from a previous work is used to collect signals from the brain with a 3D printed helmet where 5 electrodes are fastened on. These electrodes are in contact with the sensory-motor areas of the brain. An algorithm exploration tool is created with MATLAB and helps us find the combination of algorithms giving the best classification results, ie. detecting the correct movement. Then a real-time application is developed and proves to be functional by offering good results, with a real-time classification accuracy of 95% in ideal conditions and other remarkable performances.