[Eye movement detection with the aid of the oculogram in shifting the gaze]

Eye saccades are accompanied by changes of ocular electric potential. The sign of these changes involves a function of electrode location and eye movement direction while an indicator of the rotation angle is the amplitude. Based on the spatial distribution of the ocular potential one can solve an inverse problem and recover eye movement trajectory to be used for on-line computer control. To achieve this a system has to be able to place a cursor to a point on a screen corresponding to the current gaze direction of the user. We used four electrodes, two inferior and two lateral around the ocular depths. Inferior electrodes were used for estimation of the vertical gaze shift component and the lateral electrodes for estimation of the horizontal component. We detected and processed saccadic unipolar potential changes whose morphology resembles that of the step function. Detection and processing was performed using our proprietary multistage filter applied to the 4-channel data. The output of this filter was used to compute eye rotation parameters. Characteristic potential changes during the spontaneous blinks were identified and excluded from processing. Voluntary winks were used to mimic mouse clicks. In the beginning, our subjects went through a calibration stage during which they had to follow the cursor in eight basic directions. Using the calibration results the inverse problem was solved, i. e. based on the spatial distribution of ocular potential we computed screen position coordinates corresponding to the gaze direction. The presented technique belongs to the class of brain-computer interfaces. In addition, this work led us to a set of interesting observations regarding the characteristic patterns of eye movements. For instance, we found that just prior to a saccade EOG demonstrates short negative potential of 10-15 ms duration. We have also observed that with age the saccade amplitude decreases. Interestingly, when the gaze is shifted to the left, the left eye deviation significantly exceeds that of the right eye but the right shift does not exhibit such an asymmetry. During the diagonal shifts (bottom-right, top-left) the right eye skew exceeds that of the left one and the situation reverses for the two complimentary directions. We have observed no significant differences in eye coordination due to the subject gender.