This paper presents an algorithm for the automatic classification of fast and slow phases during nystagmic eye movements, using a neural network. When a patient is presented with a nonstationary visual field, the resulting eye movements may be used to determine valuable clinical information about patients with vertigo and balance disorders. When the nonstationary visual field is induced by sinusoidally rotating the patient in a chair, the eye movements—collectively referred to as nystagmus—typically consist of short, high-velocity movements (fast phases) which are in the direction of the stimulus and longer, low-velocity movements (slow phases) which are in the direction opposite to that of the stimulus. The slow phases are produced to compensate for the moving visual field. By extrapolating over the fast-phase segments, the slow-phase segments can be pieced together to form a slow-phase response. When the stimulus is sinusoidal, the slow phase response is also sinusoidal and the magnitude and phase relationships between the stimulus and response may be used to help identify the source of the patient's disorder. Thus, the ability to accurately reconstruct the response from the slow-phase segments is extremely important. This, in turn, necessitates the ability to accurately determine the locations of the fast and slow phases of nystagmus. For the neural network used here, the optimal input feature set and number of hidden units are determined, along with the necessary preprocessing of the network inputs and the postprocessing of the network output data. It is also shown that an effective error-correction algorithm can be applied to the outputs of the neural network to improve its classification ability. Finally, results are presented for the performance of the network on independent sets of test data. The classifications obtained from the neural network when applied to the test data are much more accurate than those obtained using two current classifiers: an algorithm proposed by Wall and Black and an algorithm proposed by Jell, Turnipseed and Guedry
Read full abstract