David Perruchoud

David Perruchoud

Faculty Mentor: Sydney Cash, MD, PhD
Cortical Physiology Laboratory, Massachusetts General Hospital
David Perruchoud

Project Summary: This project takes place in one of the neurology laboratories of the Massachusetts General Hospital. It aims at the understanding of the neural processing of human language production, and may be involved in the future development of a more intuitive and natural brain-computer interface for speech generation. Eventually, the new brain-computer communication approach developed by such a system could revolutionize the interface of any electronic device, by silently ordering actions or dictating text, but a more imminent application of this research may improve the quality of life of completely paralyzed patients (for ex. Locked-In Syndrome or advanced Amyotrophic Lateral Sclerosis) by providing them with a quicker and much more natural way to communicate with their family, friends or hospital staff.
This project investigated the feasibility to decode semantic information out of non-invasive brain recordings collected with electroencephalogram (EEG) and magnetoencephalogram (MEG) techniques. These data are recorded during spontaneous covert speech, that consists of silent words generation without articulation nor other motor activities; in easier words, it simply consists of saying words in your head. These data are then analyzed to recover the exact time of word onset and to decode their semantic category (determining whether the word represented for example an animal, a tool, or a person).
This study differs from most of the existing approaches that currently try to decode the different syllables or sounds of a word (phonological decoding) instead of its signification (semantic decoding). It also relies on EEG and MEG, two widely used non-invasive brain recording techniques, while most of the other studies requires heavy surgical procedures (such as electrocorticogram, microelectrode arrays or neurotrophic electrodes) and therefore prevent the evaluation of its performance on a significant pool of subjects.

Fellowships

Fellowship Year