Purpose
The electro-oculogram (EOG) is a diagnostic tool measuring components with different time courses (fast and slow oscillations, FO and SO) of light induced changes in the electrical resting potential across the retinal pigment epithelium (RPE). Here we present two algorithms for automated evaluation of SO and FO EOG, a comparison with manual evaluation, and normative data.
Methods
SO and FO EOG were recorded in healthy volunteers (SO: male/female n = 24/28, FO: n = 13/22; age: 13 - 69 yrs) according to ISCEV standards using a Diagnosys Espion e².<br /> SO amplitudes were determined by a square-wave fit (FIT), built-in analysis (RMS), and manual cursor placement (MAN). Subsequently, dark trough and light peak were determined by a cubic polynomial fit, and the Arden ratio was calculated.<br /> FO were evaluated using automated marker placement (AUT), and by fitting a sine wave to the envelope generated with an ADSR filter (ENV). Subsequently, the peak-/trough ratio was calculated.<br /> Norms were stratified for age and gender. The study was approved by the local ethics committee and performed in accordance to the Declaration of Helsinki.
Results
Manual and automated evaluation did not show significant differences regarding the Arden ratio (MAN / FIT: p = .227; MAN / RMS: p = .096) (Fig. 2b). Normative values for the Arden ratio (median (5th-95th percentile)): FIT: 234 % (167 % - 381 %); MAN: 228 % (167 % - 366 %); RMS: 217 % (164 % - 365 %). Fig. 2a lists additional values for dark trough and light peak. The Arden ratio does not correlate with the gender or age, except in manual evaluation (Fig. 2c).<br /> Peak/trough ratios determined with AUT and ENV are significantly different (p = < .0001). Normative ranges for FO peak/trough ratio (mean ± sd): ENV: 1.21 ± 0.08; AUT: 1.76 ± 0.65. There is no correlation between ratio and age (p = .281) or gender (p = .868).
Conclusions
Automated analysis of slow and fast oscillations EOG avoids time-consuming manual evaluation. Manual and automated approaches of SO show no significant difference. Automated evaluation of FO is significantly different to built-in evaluation.