By Topic

Real-time decoding algorithm in brain machine interfaces

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

7 Author(s)
Xiaoxiang Zheng ; Biomed. Eng. Dept., Zhejiang Univ., Hangzhou, China ; Qiaosheng Zhang ; Shaomin Zhang ; Yi Yu
more authors

The technology of invasive Brain-Computer Interfaces (BCIs) has been developed in last decades, for its possibility to restore motor function of the disabilities. The main task of BCI system is to translate the cortical neural activities into commands of direct brain-controlled prosthetic devices. To study how cortical signals simultaneously recorded from primary motor cortex (M1) neurons were used for external devices control, invasive brain-machine interfaces in rat were investigated. In these experiments, rats were trained to control a relay to obtain water by pressing a lever over a pressure threshold. Microwire array was implanted in rat's motor cortex to record neural activities, and pressure was recorded by a pressure sensor. After spike detecting and sorting, totally 22-58 neurons were found in all 15 channels per rat (except the reference electrode in the array). To compute the firing rate of individual unit, the numbers of spikes in a time bin (Δt =100ms) were counted. Meanwhile, the pressure signal was also computed into bin size (Δt =100ms), by averaging pressure value in the bin period. After that some decoders were designed to make mapping between neural activities and pressure, such as Optimum Linear Estimation(OLE), Kalman filter (KF), and Kernel-Based Membership combined with Curve Fitting algorithm (KMCF). These decoders can be easily learned using a few minutes of training data and provides real-time estimates of hand position every 100ms given the firing rates of neurons in motor cortex. The performances of these decoders were evaluated by Correlation Coefficient (CC) between real and predictive pressure value. The results showed that the KMCF decoder performed best in these experiment, which had higher CC than others (CC=0.94). The higher CC indicated that the activity of motor cortex (M1) neurons can be used for detection of the corresponding movement states and estimation of continuous kinematic parameters. The rats could use - - their neural activities to directly control the relay and successfully get rewards after about one week of training. In further, with the development of the motor-related BCIs technology, BCIs will be possibly used to motor function restoration of paralytics and greatly rehabilitate their capacity of life. Furthermore, the experiment results provide insights into the nature of the neural coding of movement.

Published in:

Complex Medical Engineering (CME), 2010 IEEE/ICME International Conference on

Date of Conference:

13-15 July 2010