In the analysis of optical imaging of functional brain

the traditional difference image(DI) method can not remove the vascular artifact effectively due to the existence of low frequency vibration and the wrong parameters set.In order to resolve this problem

we first study the distribution of the 0.1Hz low frequency vibration in the brain

then according to the vibration characteristic

present a new low frequency vibration difference image(LFVDI) method.The LFVDI method gets the differential maps with the help of the low frequency vibration characteristic

so it can remove the vascular artifact sharply

Compared to the DI method

the LFVDI method increases the detection performance and robustness greatly.Further

in our study

we find though in the high SNR

the DI and LFVDI methods can not detect the activity maps accurately because of the impact of phase of 0.1Hz low frequency vibration and the incorrect parameters set.So in this circumstance

we design a select strategy for the pix time series

and using this select strategy

the nonparametric temporal decorrelation source separation(TDSEP) algorithm is introduced into the optical imaging of functional brain successfully

the simulation results show that the detection performance can be improved greatly.