Purchase this article with an account.
. Vaegan, J. Nilsson, T. Wright; Transfer Functions for Rescaling Electroretinograms (ERGs) Recorded With Different Electrodes: Theory and Examples. Invest. Ophthalmol. Vis. Sci. 2008;49(13):2225. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To show, with three popular ERG electrodes, how to scale ERGs obtained with any electrode against those which would occur with another electrode.
Simultaneous bilateral ERGs and PERGs were recorded from a wide spectrum of normal persons and patients with symmetrical bilateral disease. Subjects had a DTL fiber in one eye, chosen at random and either a Burian Allen (BA) or a gold foil (GF) electrode in the other eye (N=15 and 23 respectively). To increase the correlation, components with the widest possible range of amplitudes and times to peak were included by recording from many conditions beyond the ISCEV minimum standard. The means and standard deviations (s.d.s) of the amplitude and time to peak of each component with each electrode, and the two regression equations, constrained to pass through zero of their relationship (i.e. DTL=bGF and GF=cDTL) were calculated. The ‘true’ exponents were estimated by averaging one exponent with the inverse of the other (i.e. (b+1/c)/2). The exponents for the two electrodes not directly compared were estimated algebraically. The table was further expanded to include data comparing BAs to trans-corneal DTLs under a bandage contact lens, available from the literature.
DTL had 0.64xBA amplitude and 1.2xBA s.d. DTL had 0.86xGF amplitude and 0.76xGF s.d. Trans-corneal DTL had 0.99xBA amplitude and 0.84xBA s.d. Peak time measures hardly varied across electrodes. It can be assumed that all time exponents are 1.00 and remaining deviations from it reflect the residual error. In log/log plots small components were on the same straight line. Trans corneal DTL had better SNR than BA, and sub palpebral DTL than GF because s.d. decreased more than amplitude.
This method makes it possible to estimate the relative SNRs of electrodes by also scaling variance. Laboratories can then adopt a new electrode and still make use of previously collected normal values. Normal values can be shared between clinics that use different electrodes, and results from individual cases, recorded elsewhere, can be evaluated against local norms, but only if all other conditions are identical. New laboratories will be able to start work rapidly, without establishing their own normative data base if they meet this criteria. This should encourage the use of strict standards.
This PDF is available to Subscribers Only