Nonlinear approaches for separation of slow and fast phase nystagmus signals

Central nervous system (CNS) disorders and vestibulo-ocular reflex (VOR) degradation contribute to the loss of orientation and balance, typically causing the illusion of motion. VOR degradation may be caused by illnesses or common prescription medications, and this degradation may have serious repercussions in affecting one's ability to walk, maintain balance, or track moving objects. The two current VOR analysis techniques are applied to nystagmus signals (recordings of eye position during standardized tests) to isolate the two neuronal pathways used. These techniques are only partially successful because the VOR signal is passed through several filtering layers, to isolate the nystagmus signal, leaving it severely altered. In this paper, several alternate approaches are introduced: nonlinear robust ordering procedures, mathematical morphology, and wavelet filtering techniques are demonstrated. All approaches were as successful as the current techniques in isolating data from one neuronal pathway. Techniques for each of the three were also developed which would identify data from the second neuronal pathway as well. The nonlinear approach, which works using raw position data, bypasses layers of filtering. A robust technique is used to identify the points at which the data changes from that due to one neuronal pathway to the other, and the end result is obtained by concatenating the respective segments. The other two approaches work with velocity data obtained by differentiation, and because they are applied to non-original data, their results may be open to the possibility of error. Additionally, the two most recent approaches in the literature (using order statistic filters and fuzzy logic) have been implemented for a comparison of results.