Abstract
Abstract: :
Purpose:To test the proposal that the signal–to–noise ratio (SNR) of the multifocal visual evoked potential (mfVEP) can be increased using a combination of higher order kernels (i.e. 2.2K and 4K) and principal component analysis (PCA)[1]. Methods: Monocular mfVEPs were recorded from both eyes of 100 normal subjects [2] using VERIS software (EDI) and a 60–sector, dartboard array. Two 7–minute recordings were obtained from each eye using effectively 6 channels of recording [1]. The SNR of the best channel was used as the measure of mfVEP amplitude. First, the first 4 principal components (PCs) were obtained from each of the kernels. Then the combined PCs were obtained using the bases of the 2.1K PCs and the weighted average of the coefficients of the PCs of all three kernels. For each subject, the weights were determined by maximizing the t–statistic between the signals and the noises. New local records were reconstructed by combining weighted PCs and then compared to the 2.1K records. The data for the first run (1run) and the average data from both runs (2run) were analyzed. False positives were estimated using a window of the response containing only noise. Results:The mean SNR was 2.4 (1run) and 3.1 (2run) for the PCA–kernel (PCAK) method compared to 2.0 (1run) and 2.4 (2run) for the standard method (STD). Using the 2.5 percentile, the median false alarm rates (the percentage of noise windows falling above the local 2.5 percentile) for the interocular test were 11.5% (1run) and 3% (2run) for PCAK compared to 18% (1run) and 5.5% (2run) for STD. For the monocular test, the rates were 3% (1run) and 0% (2run) for PCAK compared to 9.2% (1run) and 1% (2run) STD. Conclusions:The mean SNR was significantly larger for the PCAK method compared to the standard method. In fact, the 1run SNR distribution was essentially the same as the 2run STD method. As measured with the noise window, the false alarm rate was also lower for the PCAK method, but the 1run results were not as good as the 2 run STD results. [1] Zhang& Hood (2004) JOV. [2] B. Fortune et al. in press DOOP.
Keywords: electrophysiology: clinical • visual fields