I am interested in elucidating the neuronal system behavior through novel signal processing and computer modeling techniques. Some of my previous activities include early detection of seizures, neural computational modeling, analysis of stochastic resonance and network synchronization in neuronal networks, as well as feedback control strategies for prosthetic devices.  My current work mainly deals with the development of brain-computer interface technologies to improve usability.

This work addresses the need to improve the user acceptability of the rehabilitation technology by making the user control more intuitive. A nonlinear model predictive controller algorithm was implemented to modulate the grasping force and slippage of a prosthetic hand. We have also developed a brain-computer interface (BCI) technology to recognize the user’s intention as well as imagined movements.

Our group uses evoked response potentials elicited in rapid serial visual presentation tasks to assess how text comprehension, cognitive loads, and fatigue can be used as markers to improve error-correction capability in brain-computer interface applications. The aim of this research is to develop communication protocols where the control parameters can be adapted based on the mental states of the users, by identifying the appropriate behavioral and electrophysiological markers.

This work deals with the anticipation and characterization of abnormal neural activities. Our group has explored different supervised and unsupervised pattern recognition techniques to classify brain signals undergoing state transitions into seizure episodes. We have also investigated how electrical field orientation and extrinsic noise intensity can help promote the stochastic resonance phenomenon in neuronal networks.

The oscillator-based model was developed to represent the state transitions of in vitro hippocampal slice data. Each neural unit can be mathematically described as an oscillator capable of generating regular action potential spike trains without external inputs, or a threshold-based spiking unit. The output of the oscillator model was used as a stimulation signal and its effectiveness to suppress seizures has been successfully demonstrated.

updated 11/23/2019