Noise-Related Variance of Functional Networks

Hu Cheng¹, Aina Puce¹

A simple method is proposed to estimate the variance of resting state functional network originated from residual noises in the MRI signal. The variance from noise is compared with the total variance between networks from different time periods. The results suggest that a substantial amount of variance of the functional network comes from the intrinsic noise that is not coupled with the coherences of different brain regions. Sampling more time points can effectively reduce the noise-related variance. Another effect of residual noise is reduction of the correlation coefficient, which can be estimated from the variance of network.