Classification of infantile nystagmus waveforms.

Classification of infantile nystagmus waveforms is an important problem because the characteristic waveforms can be used to distinguish between infantile and acquired nystagmus. A clear description of the nystagmus is also a necessary first stage in understanding its origin. Currently infantile nystagmus waveforms are classified into at least 12 different types. In this study we analyse a database of nystagmus recordings in order to investigate if this classification can be simplified. Application of principal components analysis revealed that 96.9% of the variance of the waveforms is described by a linear sum of two component waveforms. The components consist of sawtooth and pseudocycloid waveforms that account for 78.7% and 18.2% of the variance respectively for the most common single cycle waveforms. This simplified description of infantile nystagmus highlights the importance of identifying the origin of the jerk component and its synchronisation with the pseudocycloid component for the characterisation and treatment of the nystagmus.