|STATISTICAL ANALYSIS OF THE PATTERN ELECTRORETINOGRAM (PERG) SIGNAL|
|KAZIMIERZ SZLACHTA, KRZYSZTOF PENKALA, ANDRZEJ BRYKALSKI, WOJCIECH LUBIŃSKI|
Pattern electroretinogram (PERG) is a bioelectric signal recorded from the human retina with a corneal electrode, when a specific optical stimulus is applied to the eye. PERG recordings are very important in diagnosis of many diseases of the retina as well as optic nerve - e.g. glaucoma. In the paper a part of the authors' research project is presented - application of statistical methods for determining certain features of the PERG waveforms. For the purpose of the present study, 60 normal PERG waveforms and 47 recordings typical of some previously diagnosed retinal and optic nerve diseases were chosen. They were obtained in two age groups and recorded in accordance with the ISCEV guidelines. Classification of these signals using discriminant analysis and logistic regression was performed. It resulted in obtaining predictive, quantitative models, which allow fast evaluation of the functional state of the retina.
|Key Words:||pattern electroretinogram, PERG, statistical classification, discriminant analysis|