AUTHOR=Bouton Chad , Bhagat Nikunj , Chandrasekaran Santosh , Herrero Jose , Markowitz Noah , Espinal Elizabeth , Kim Joo-won , Ramdeo Richard , Xu Junqian , Glasser Matthew F. , Bickel Stephan , Mehta Ashesh 

TITLE=Decoding Neural Activity in Sulcal and White Matter Areas of the Brain to Accurately Predict Individual Finger Movement and Tactile Stimuli of the Human Hand

JOURNAL=Frontiers in Neuroscience

VOLUME=Volume 15 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2021.699631

DOI=10.3389/fnins.2021.699631

ISSN=1662-453X

ABSTRACT=Millions of people suffer motor or sensory impairment due to stroke, spinal cord injury, and many other conditions. A brain-computer interface (BCI) offers regained independence, but one of the challenges currently facing BCI technology is surgical risk. Stereoelectroencephalography (SEEG) reduces risk and gives access to sulcal and white matter areas of the brain, but SEEG has not been widely studied in the BCI field. We therefore investigated the viability of using SEEG electrodes in a BCI for recording and decoding neural signals related to hand movement and the sense of touch and compared its performance to electrocorticography electrodes (ECoG). Initial poor decoding performance led to the development of a feature selection method based on temporal autocorrelation, a repeatability metric. This identified stable features (important for accurate decoding), along with deep learning methods, to automatically classify various motor and sensory events for individual fingers with high accuracy. Repeating features were found in sulcal, gyral, and white matter areas across a wide frequency range. These findings motivated the use of long short-term memory (LSTM) recurrent neural networks (RNNs). Combining temporal autocorrelation-based feature selection with LSTM yielded decoding accuracies of up to 92.04 +/- 1.51% for hand movements, up to 91.69 +/- 0.49% for individual finger movements, and up to 80.64 +/- 1.64% for focal tactile stimuli to the finger pads and palm while using few SEEG electrodes. These findings may lead to a new class of minimally invasive BCI systems in the future for a wide range of conditions.