Multivariate statistical analysis of average visual evoked potentials in functional amblyopia

Abstract

This is a study of the visual evoked potentials of fifty-six children with unilateral functional amblyopia. By means of multivariate statistical analysis (correlation matrix, factorial analysis of correspondence, step-by-step discriminant analysis) the following points have been shown. 1. A. If the average curves summarizing the individual visual evoked potentials associated with the normal and the abnormal (amblyopic) eyes are considered: (1) There is congruence between the two curves; the abnormal curve begins later, remains smaller and ends earlier than the normal curve, which completely encloses it. (2) The best correlations between these curves occur at points with different latencies, comparable in relation to the form of each curve. 2. B. Considering individual evoked potentials: Their latency is much greater and their amplitude much smaller as visual acuity, both near and distant, becomes lower, and the fixation point is farther from the fovea. (2) If, in individual cases, one tries to predict the result of functional reeducation in terms of the visual evoked potential it is found that: (a) the maximal amplitude alone gives a correct estimate in 60% of cases; (b) discriminant analysis of the whole visual evoked potential gives a correct estimate in 86% of the cases, the same as for the clinical findings; (c) the conbination of the visual evoked potential and clinical findings gives the correct prognosis in 96% of cases. This: (1) confirms the changes in visual evoked potentials with functional amblyopia; (2) shows the value of multivariate methods of analysis for comparison of the curves; (3) provides a useful method of clinical prognosis.