Reduction of light source noise from optical intrinsic signals of mouse neocortex by using independent component analysis

Because the optical intrinsic signal (OIS) of the brain is very weak, noise reduction is essential. Independent component analysis (ICA) is widely used for noise reduction. However, the applicability of ICA to the reduction of light source (LS) noise has not been discussed in detail. In addition, determining the proper number of independent components (ICs) for decomposition is very important to a reasonable classification of the ICs. In this study, we considered the applicability of ICA to LS noise reduction by modeling the impact of LS noise on OIS data. We propose a method for determining the number of ICs that uses the power spectral density of LS noise. To evaluate its usefulness, the method was applied to real OIS data of a mouse´s cerebral cortex.