Control of prosthetic limbs from decoded brain signals

Summary This research will restore mobility to patients who suffer from paralysis. We aim to create a device, known as a brain-machine interface, which is an artificial communication path from the brain that bypasses an injury, such as a damaged spinal cord or stroke. The interface will decode a user’s intent and act upon it. Decoders will use physiological principals and state-of-the-art machine learning methods. We will test a user’s ability to control an artificial limb using decoded brain activity.

Project Details

This research will restore mobility to patients who suffer from paralysis. We aim to create a device, known as a brain-machine interface, which is an artificial communication path from the brain that bypasses an injury, such as a damaged spinal cord or stroke. The interface will decode a user’s intent and act upon it. Decoders will use physiological principals and state-of-the-art machine learning methods. We will test a user’s ability to control an artificial limb using decoded brain activity.

This project will demonstrate proof of concept of the clinical viability of a device that will restore mobility to the millions of people worldwide. The device, known a brain-machine interface, will serve as an artificial communication channel from the brain that bypasses damaged tissue, such lesions caused by stroke or spinal cord injury. This interface will enable computer control by decoding the electrical activity of the brain, allowing communication with robotic prostheses, enabling people to reconnect with the physical world.

Despite the striking demonstrations of brain-machine interfaces for driving prosthetic devices, this technology has not been translated to the clinic. The major reason for this is that the electrode systems that capture the neural signals are unreliable. Consequently, the lifespan of these devices is limited.

We have recently solved the reliability problem and published two approaches for the successful decoding of the local field potentials, which are more stable than standard approaches. We have established methods that are based physiological principals and state-of-the-art machine learning approaches that solve complexity issues of local field potential decoders. Furthermore, we also have unequivocal evidence that local field potentials are reliable for long-term continuous recording.

In this project, we will directly test our brain-machine interface designs in humans who have subdural electrodes placed on the surface of their brains for epilepsy surgery purposes. We will assess the ability of these subjects to control a robotic arm in real time using decoded intracranial EEG signals. There is a strong need for brain-machine interfaces to restore mobility to people living with paralysis. We have a wonderful opportunity to provide freedom to millions, to advance medical technology in Australia, and to push the boundaries of science and advance our knowledge of the human brain.