Visual evoked potential-based acuity assessment in normal vision, artificially degraded vision, and in patients

To assess visual acuity (VA) objectively using visual evoked potentials (VEPs), avoiding subjective trace evaluation and providing an acuity estimate with associated confidence limits.

40 normal subjects and 24 patients (with corneal and retinal diseases, decimal VA range 0.15-1.1 (= 0.8(logMAR) to -0.04(logMAR))) participated in the study. Checkerboard stimuli with six check sizes covering 0.05-0.4 degrees (or 0.09-0.8 degrees for visual acuities below 0.35 (= 0.46(logMAR)) were presented in brief-onset mode (40 ms on, 93 ms off) at 7.5 Hz. In normal subjects, the stimuli were also optically degraded by frosted occluders resulting in a decimal VA range of 0.13-2.8 (= 0.9(logMAR) to -0.45(logMAR)). Altogether, 108 steady-state VEPs were recorded with a Laplacian montage (2xOz-(RO+LO)). Fourier analysis yielded the magnitude (A) at the stimulus frequency, and the average of the two neighboring frequencies as noise estimate (N). A and N determine the significance level p of the response, and from their ratio the non-noise-contaminated response (A*) can be calculated. Tuning curves were obtained by plotting A* vs the dominant spatial frequency of the corresponding checkerboard. A fully automatic algorithm used the significance level (p<5%) and A* to automatically select an appropriate region in the high spatial-frequency range on which a linear regression was performed, yielding a zero-amplitude extrapolated spatial frequency SF0. Subjective VA was obtained with the automated "Freiburg Acuity Test".

The brief-onset presentation evoked high VEP amplitudes; however, many tuning curves displayed the well-known "notch" at intermediate check sizes. The fully automated analysis algorithm succeeded in 107 of 108 cases and effectively ignored the notch, if present. The relation between logVA and log(SF0) was a constant factor throughout the range tested: logVA = log(SF0)/17.6 cpd. In more than 95% of all cases, the acuity predicted from SF0 coincided within a factor of two (up and down, or +/-0.3 logMAR) with subjective VA with a coefficient of correlation of 0.90.

The fully automated analysis avoided subjective problems in peak-trough assessment. The results provide quantitative limits to assess patients with possible malingering.