fMRI Preprocessing

A series of preprocessing steps were conducted in SPM8 [38] for effective data analysis:

The first five volumes were removed from the data for further processing to ensure magnetization equilibrium.

In order to compensate for differences in the time of slice acquisition, slice-timing correction was performed. Furthermore, the time corresponding to the first slice was chosen to be the reference.

To compensate for bulk head movements, realignment (motion correction) was done.

To map the functional and subject-matched structural images to each other, co-registration was performed.

By using a voxel size of 2 × 2 × 2 mm³, spatial normalization to the Montreal Neurological Institute (MNI) standard space was performed. The product of this step was images with 79 × 95 × 68 spatial dimensions.

The fMRI data were smoothed in order to increase the signal to noise ratio (SNR). Gaussian filters are commonly used to smooth images [39, 40]. In the present study, for spatial smoothing, Gaussian kernel of 8 × 8 × 8 full-width half-maximum (FWHM) mm³ was utilized.

Numerical normalization was done. In this step, values of each fMRI data need to be between zero and one.