To understand the neural basis of speech encoding in human motor cortex and to develop algorithms to decode intended speech using an intracortical brain computer interface (iBCI).

For people with ALS, locked-in syndrome, and other conditions that cause quadriparesis and severe dysarthria/anarthria, the loss of fluent communication is highly disabling. Previously, iBCIs have allowed people with paralysis to type and write individual letters and decode phonemes; an ECoG BCI has also been used to decode a vocabulary of 50 spoken words. Here we describe our efforts to decode intended speech using an iBCI; with signals from populations of single neurons and local field potentials, we anticipate this approach will allow accurate and robust decoding.  

Research is conducted with permission under an Investigational Device Exemption from US FDA and the MGH IRB. A 37-year-old with quadriplegia from a spinal cord injury enrolled in the BrainGate clinical trial had two 96-channel microelectrode arrays placed chronically in dominant precentral gyrus. During recording sessions at his home, the participant read pseudo-randomly presented words and sentences while neural activity was recorded.

Principal component analysis identified the 100-dimensional feature space capturing the greatest variance across the speech task. A support vector machine (SVM)-based decoder was used to distinguish between each spoken word and epochs of silence. In cross-validated analyses, the SVM correctly distinguished epochs of speech from silence in 96.6% of 2440ms epochs spanning 1176 spoken words over 3220 seconds of word reading. Individual articulatory elements of speech were also well discriminated, with clustering of words observed most strongly on initial consonant sound. 

In this preliminary study, iBCIs can be used to detect intended speech directly from motor cortex. Ongoing work is focused on enhancing and refining decoding algorithms to restore fluent speech-based communication to people with severe dysarthria and anarthria.