Brain-Machine Interfaces—Taking the Next Step Towards Translational Devices

Aphasia is defined as the impairment of speech as a result of acquired brain injury, rapidly hindering communication skills in the patient. Current therapeutical approaches are complex and rely on compensatory strategies to regain the lost communicating abilities. <br/><br/>Based on the concept proved by Khodagoly et al, multielectrode arrays (MEA) using electrodes as small as 10x10 μm² are sufficient to capture both local field potential and putative single neuron activity when used for electrocorticography, causing minimal tissue damage. To date, MEAs have predominately been used for investigating human neural network mechanisms and better understanding of brain disorders. Nonetheless, better interpreting these mechanisms carries the potential to open an entire new medical field to use brain signal recordings for healthcare applications and personalized medicine.<br/><br/>In this work, we present the first translational Neurogrid (ultraconformal ECoG multielectrode array) that will be used for the chronic study of cortical neural signals and its decodification into conversation. By complying with European medical regulation in the development of the device, we aim to increase the duration of the implantation in awake human studies, to better understand the potential of this technology with respect to long-term human-machine interfaces. Integrating state-of-the-art machine learning algorithms to the brain signals recorded by the Neurogrid, we aim to decode neural activity in patients suffering from aphasia and translate it into speech.<br/><br/>The combination of these powerful technologies can pave the way for brain-machine interfaces to become powerful clinical tools that increase the wellbeing of patients with neurological disorders. However, to champion the way of this technology into a final clinical product, it should still undergo multiple intermediate steps to prove its safety and reliability.