Purchase this article with an account.
Yao Chen, Xinyu Chai, Yiliang Lu, Pengjia Cao, Liming Li, Qiushi Ren; Independent Component Analysis Can Remove Cortical Artefacts Elicited by Electrical Optic Nerve Stimulation. Invest. Ophthalmol. Vis. Sci. 2012;53(14):2472. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Electrically evoked potentials (EEPs) can be elicited by one or more biphasic current pulses delivered to optic nerve (ON) from penetrating electrodes. Comparing EEPs with visually evoked potentials (VEPs), we can assess the effects of varying stimulus parameters and evaluate the functionality of visual prosthesis. However, EEPs recorded from the visual cortex usually contain large stimulus artefacts caused by instantaneous electronic current spread through the brain tissue. These stimulus artefacts contaminate the EEP waveform and often make subsequent analysis of the underlying neural responses difficult. Therefore it is important to eliminate artefacts and reconstruct the cortical potentials by signal processing algorithms.
We recorded multi-channel EEPs from visual cortices of five rabbits in response to ON electrically stimulation with four different stimulus parameters. Independent component analysis (ICA) was developed and applied to remove electrical stimulation-induced artefacts and reconstruct cortical responses. ON action potentials were then blocked by lidocaine in order to acquire cortical signals only including stimulus artefacts. Correlation between reconstructed artefacts by ICA and artefacts recorded after blocking the ON was analysed to estimate artefact removal performance.
Mean values of the calculated correlation coefficients under four different ON electrical stimulation conditions are larger than 0.9, and there is no significant differences among the four conditions (one-way ANOVA, p = 0.11). This indicates successful removal of artefacts from electrical stimulation by the ICA method. The same correlation analysis as above was repeated using 3 × 3, 2 × 3 and 1 × 3 recording arrays to determine if there was an effect related to number of recording channels. Our results show that, when applying ICA separation, the lowest correlation coefficients were obtained with a 1 x 3 array and these values were significantly lower than those obtained with 3 x 3 and 2 x 3 arrays (which were not significantly different from each other, one-way ANOVA, p = 0.01).
ICA technique has potential applications in studies designed to optimize the electrical stimulation parameters used by visual prostheses.
This PDF is available to Subscribers Only