Analyzing ERG Oscillatory Potential in Mouse Using Hilbert-Huang Transform

Abstract

Purpose:: Fourier transform has been applied in ERG signal analysis to extract signal components of particular frequencies such as the oscillatory potentials (OPs). However, conventional Fourier analysis can not derive time-dependent frequency evolution, which may provide valuable information regarding the signal originations. Hilbert-Huang transform is a newly developed algorithm and proven to be effective in analyzing nonstationary and nonlinear signals. We have applied this method to analyze mouse ERGs that present prominent OPs.

Methods:: Dark- and light-adapted electroretinograms (ERG) were obtained in three mouse strains: wild-type C57BL/6J mouse, cone photoreceptor function loss 1 (cpfl1) mouse, and rhodopsin knockout (rho^-/-) mouse. To apply Hilbert-Huang transform, the time domain empirical mode decomposition (EMD) algorithm was used to break down the ERG signals into a series of intrinsic mode functions (IMFs). Then Hilbert transform was applied to derive the time-dependent signal frequencies and amplitudes.

Results:: Empirical mode decomposition of mouse ERG usually produced 6-8 intrinsic mode functions. The first 2~3 IMFs contained high frequency signals and were coincident with the temporal locations of OPs. The first IMF has the highest frequency. In dark-adapted C57BL/6J mouse, the first IMF appeared at -3.35 log cd/m² with a peak amplitude of 15.6µV at 92Hz. It increased to 131.9µV at 110Hz under light intensity of 0.65 log cd/m². In dark-adapted cpfl1 mouse, the first IMF appeared at -1.85 log cd/m² with highest amplitude of 11.7µV at 120Hz. It increased to 90.4µV at 120Hz under light intensity of 0.65 log cd/m². In light-adapted measurements, the C57BL/6J mouse showed 29.3µV peak amplitude at 75Hz at 0.65 log cd/m² in the first IMF, which was similar to that obtained from rho^-/- mouse (26.9µV, 75Hz).

Conclusions:: We have applied Hilbert-Huang transform in ERG signal analysis. An advantage of this time domain method is that it obtains time dependent frequency information. The transform may provide a novel method to decipher the ERG signals.

Keywords: electroretinography: non-clinical • photoreceptors