Objective Methods for the Detection of Pattern Electroretinograms (PERG)

摘要

Pattern electroretinograms (PERG) are evoked responses recorded from electrodes on the cornea in response to stimulation by a changing visual pattern image. Conventional analysis of these signals relies on subjective interpretation by a clinical expert. We propose a method that is automated and objective, using the bootstrap technique to determine the statistical significance of parameters that quantify the strength and/or shape of the response. The methods were tested on 54 PERG recordings (including 18 no-stimulus controls). The results show high sensitivity of the methods, with detection of the response in almost all cases after only 20 stimuli, which compares favourably to magnitude squared coherence, an alternative conventional statistical test. False-positive rates were at the expected 5% level. The proposed bootstrap method for the detection of PERGS method is very flexible in the choice of signal features to be tested, and can be applied on other evoked potentials.