Abstract
Purpose: :
Pattern electroretinograms (PERG) obtained at stimulation rates higher than 6 rev/s result in steady-state responses which are hypothesized to arise from the convolution of overlapping transient responses generated at that rate. Previously, we developed a deconvolution algorithm to extract transient responses from steady state responses obtained using slightly jittered stimulus sequences. In this study this generation hypothesis is tested by generating synthetic steady state responses from extracted transient responses and comparing them to conventional steady state responses.
Methods: :
A specially constructed, fast responding LED visual display was used to deliver pattern reversals. Jittered quasi-steady-state PERGs (qss-PERG) and true steady state PERGs (ss-PERG) were obtained using lower eyelid electrodes at five rates (6.9, 10.9, 15.4, 17.4, 26.5 rev/s) for each subject. Additionally transient PERGs (t-PERG) at 2.17 rev/s were recorded for comparison. For each rate extracted t-PERGs were convolved to construct a synthetic ss-PERG. The synthetic PERGs were then compared to true ss-PERGs. Correlation and phasor analyses were used to assess the validity of the generation hypothesis.
Results: :
Waveform morphology (N35, P50, N95 components) is very stable through 17.4 rev/sec, but as previously observed changes significantly at higher rates. Synthetic ss-PERGs constructed with deconvolved transient responses superimposed well (correlation > 0.90) with true ss-PERGs. Magnitude and phase analysis obtained further confirmed these findings. Synthetic PERGs constructed using 2.17 rev/s transient responses was good predictor at low rates did not predict ss-PERGs acquired at higher rates as well (correlation values dropped to 0.85 and 0.26 for 17.4 and 26.5 rev/sec respectively).
Conclusions: :
The findings of this study corroborate the hypothesis that ss-PERGs arises from the superposition of overlapping transient responses. Additionally, for all rates tested deconvolved PERGs were a better predictor of ss-PERG than low rate t-PERG. By using jittered stimulus sequences, both transient and steady state responses can be estimated at any rate and temporal adaptation effects can be revealed. Deconvolved PERG responses are advantageous as they retain temporal information (N35, P50 and N95) that is lost in steady state at higher rates.
Keywords: electroretinography: non-clinical • ganglion cells