PowerPoint Presentation - Time Series Analysis in AFNI

–60–

Sticks and TR-locked Stimuli

• h(t ) = 0 for any t not on the TR grid

• This limitation means that, using stick functions as our basis set, we can only model stimuli that are “locked” to the TR grid

H That is, stimuli/activations don’t occur at fully general times, but only occur at integer multiples of TR

• For example, suppose an activation is at t =1.7×TR

H We need to model the response at later times, such as 2×TR, 3×TR, etc., so need to model h(t ) at times such as t=(2-1.7)×TR=0.3×TR, t=1.3×TR, etc., after the stimulus

• But the stick function model doesn’t allow for such intermediate times

• or, can allow Dt for sticks to be a fraction of TR for data

• e.g., Dt = TR/2 , which implies twice as many bq parameters to cover the same time interval (time interval needed is set by hemodynamics)

•then would allow stimuli that occur on TR-grid or halfway in-between