Research

Majority of daily life tasks require the user to modulate different kinematics parameters of movement like speed, direction, trajectory, force, etc. to complete the function. If we can decode the kinematics parameters associated with imagined movement from brain signals, then the same can be used to design motor imagery-based Brain-Computer Interface systems, which can provide more naturalist control to the daily life requirements of the user. In this work, we attempt to decode the speed and direction of imagined hand movement from EEG. 

Stroke is a public health problem, which may result in motor impairment, speech and audio impairment, sensorimotor loss, cognitive decline and emotional disturbances. Post stroke rehabilitation are often necessary to improve and restore the motor skills. The patients play a passive role in conventional physical practice of rehabilitation. Motor imagery and intention based BCIs are found to enhance the engagement and motivation of stroke patients and helps to improve their motor skills by promoting neuroplasticity, which is the intrinsic ability of human brain to rewire its circuits and form newer connections. A neurofeedback BCI game is developed in the form of a ball sorting game, where the subject controlled the movement of the ball through motor imagery of their right and left hand. This research is conducted in clinical collaboration with National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore.

Electroencephalogram (EEG)-based non-invasive brain-computer interface (BCI) systems decode brain signals to establish an augmented communication and control pathway between the brain and the computer. The development of an efficient BCI system requires accurate decoding of neural activity underlying the user's intentions. This study investigates the directional tuning of EEG characteristics from the posterior parietal region, associated with bidirectional hand movement imagination or motor imagery (MI) in left and right directions.  The imagined movement directions of the chosen hand were decoded using a combination of envelope and phase features derived from parietal EEGs of both hemispheres. The proposed algorithm uses wavelets for spectral decomposition, and discriminative subject-specific subband levels are identified based on Fisher analysis of envelope and phase features. The selected features from the discriminative subband levels are used to classify left and right MI directions of the hand using a support vector machine classifier. Furthermore, the performance of the proposed algorithm is evaluated by incorporating a maximum-variance-based EEG time bin selection algorithm. With the time bin selection approach using subject-specific features, the proposed algorithm yielded an average left vs right MI direction decoding accuracy of 73.33% across 15 healthy subjects. In addition, the decoding accuracy offered by the phase features was higher than that of the envelope features, indicating the importance of phase features in MI kinematics decoding. 

The utility of EEG features in decoding the movement directions was investigated by several researchers.  However, the significance of discriminative EEG time bin for motor imagery direction decoding is a less investigated problem.  The time window that offers the highest discrimination can vary from subject to subject and trial to trial. Hence, the selection of discriminative EEG time bin is very crucial in practical deployment because it helps in improving the accuracy with reduced time samples which in turn reduces the computational complexity.  A variance-based time bin selection algorithm is proposed to select the most discriminative EEG time segment for directional classification of movement imagination.  The subsegment that maximizes the absolute difference in variances of EEG signals from C3 and C4 channels is selected as the most discriminative time bin. The discriminative EEG features carrying motor imagery (MI) directional information are extracted from the selected EEG time segment using the wavelet-common spatial pattern (WCSP) algorithm. The WCSP features are classified using Support Vector Machine classifier resulting in a cross validated classification accuracy of 71% between left versus right MI directions of 15 subjects.