lightniing Projects About Source
All projects

Temporal Processing

Temporal preprocessing accounts for differences in when neuroimaging signals are acquired.

Slice TimingDetrending3dTshift

Slice-timing correction

Echo-planar imaging (EPI) commonly acquires functional MRI (fMRI) as a sequence of two-dimensional slices. Slices within each repetition time (TR) therefore have different acquisition times, whether the order is ascending, descending, or interleaved. Slice-timing correction uses temporal interpolation, typically sinc or spline interpolation, to resample each blood-oxygen-level-dependent (BOLD) time series to a common reference time. This improves temporal alignment across slices before modeling.

Slice timing correction Sequential acquisition samples axial slices from inferior to superior at different times. Slice timing correction interpolates every slice to the middle acquisition time. Slice timing correction Aligning sequentially acquired slices to a common temporal reference 01 Sequential acquisition time superior inferior 15 25 35 45 55 65 75 start end of TR 02 After temporal interpolation time superior inferior middle timepoint 15 25 35 45 55 65 75 Faint markers show the original acquisition times; arrows show interpolation to the shared reference time.
Sequential slice acquisition and interpolation to a common reference time

lightniing accelerates AFNI's 3dTshift while preserving functional equivalence.

Slice Time CorrectionAdjusting for different slices in a 3D Volume Slice Time CorrectionAdjusting for different slices in a 3D VolumeFSL slicetimer0.72xAFNI 3dTshift1xniimath stc2.8xniimath stc (OpenMP)9.8x
Illustrative Temporal Processing benchmark figure

Benchmark source

Links