ECoG Based BCI for Control of a Robotic Arm in Three Dimensions





Keywords: electrocorticography, robotics, brain computer interface, cortex, spinal cord injury

Interactive Manuscript

Ask Questions of this Manuscript:

   



Abstract

     Non-human primates have been taught to use Brain Computer Interface (BCI) technology to control robotic arms for tasks, including self feeding. Subjects with penetrating electrodes implanted long term have been able to achieve 2D control plus click of a computer cursor.
     We propose using high density subdural electrocorticography (ECoG) electrodes as a BCI for achieving mutidimensional control.
     A subject with C4 quadrapelgia underwent MRI, fMRI and MEG to localize areas of activation for imagined motion of the right arm.
     A custom designed 28 contact ECoG grid was placed using image guidance. The subject participated in two 2 hour training sessions 6 days/week. The grid was explanted on POD 28.
     The implant surgery performed with image guidance was performed in approximately 2 hours and the subject was admitted overnight. Testing commenced POD 3. Using observation based tasks we identified discrete areas of activation. Over the testing period the subject was able to achieve a high degree of accuracy in both 2D and 3D tasks using a virtual environment and a prosthetic arm. The grid was explanted POD 28 with no signs of infection and the subject was discharged POD 1.
     This is a retrospective study.
     A subject with an ECoG based BCI achieved 3D control in less than 4 weeks.
     No complications were experienced, indicating that ECoG based BCIs are effective interfaces for robotic prosthetic control.


Acknowledgements

Project Roles: