Neuroimaging insights in age-related macular degeneration (AMD): predicting visual function via embedded video markers

Abstract

Purpose : Electroencephalographic (EEG) methods aiming to quantify visual function in age-related macular degeneration (AMD) avoid subjective behavioral responses, yet typically employ simple stimuli with low ecological validity and require patients to fixate centrally. We conducted a causal-comparative study to test the hypothesis that steady-state visual evoked potentials (SSVEPs) elicited by higher vs. lower spatial frequencies (SFs) in natural-scene videos would predict visual function in AMD.

Methods : EEG was recorded from seven human patients diagnosed with AMD (Age M = 76, SD = 3.74, 1 male) and six healthy, approximately age-matched controls (Age M = 61.33, SD = 3.27, 2 males) during free video viewing. Steerable wavelet transforms were applied to filter obliquely oriented visual information in each video frame such that high and low SFs flickered at distinct temporal frequencies (7Hz & 9Hz). Videos were consecutively streamed in random order (6 x 7.5s videos x 5 reps x 2 flicker conditions = 7.5 mins). Pre-processed EEG data were averaged for each flicker condition and subjected to fast-fourier transforms. SSVEP amplitudes were calculated as the mean amplitude across flicker frequencies and electrodes for each SF.

Results : There was substantial evidence for an interaction between the effects of SF and group on SSVEP amplitudes (SNR, BF₁₀ = 7.01), such that high SF SSVEPs were weaker for AMD patients (M = 2.66, SD = 1.71) compared with controls (M = 12.84, SD = 13.91) and low SF SSVEPs were stronger for AMD patients (M = 13.20, SD = 10.12) compared with controls (M = 5.02, SD = 2.27). The ratio (low/high SF) of SSVEPs linearly predicted behavioral visual acuity (logMAR, β = 0.74, BF₁₀ = 61.27) and contrast sensitivity scores (logCS, β = -0.61, BF₁₀ = 9.70).

Conclusions : This novel, fast, ecologically valid assessment provides a quantitative neural measure of visual function which is predictive of established behavioral measures for AMD. Ongoing research efforts focus on developing scotoma mapping applications of this method, and further data collection.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.

 

Results Overview a) Example video frame. b) SSVEPs for AMD patients and controls. c) Topographic distributions of SSVEP ratios (low/ high SFs). Electrodes used to calculate SSVEP values are highlighted (POz, PO3, PO4). d) Scatter plots showing relationships between SSVEP ratios and behavioral measures of visual function.