As an alternative to z-transformation of data, applying the per cent signal change (PSC) transformation has two advantages: the signal amplitude is not altered based on variance, which means that data containing additional noise will not undergo additional scaling (unless the noise has a strong bias on the mean, that is). Additionally, data that has been PSC transformed will yield (close to) per cent signal change beta values as result of the regression equation (given that the HRF convolution has been applied in a way that ensures that the hypothesized amplitude of a prolonged and continuous stimulation is modeled as 1).

  psctrans  - perform PSC transformation on time course
 
  FORMAT:       [psctc, pscf] = psctrans(tc [, dim [, tp]])
 
  Input fields:
 
        tc          time course data
        dim         temporal dimension (default: first non-singleton)
        tp          time points (indices of dim to use for normalization)
 
  Output fields:
 
        psctc       PSC transformed time course
        pscf        PSC transformation factor
 
  See also ztrans

The new values are computes as (100 ./ mean(T)) .* T, whereas the mean is only estimated over the selected timepoints (e.g. without known to be noise).

• applying PSC transformation on a single time-course:

  psctrans_example1.m

  ptc = psctrans(tc);

• PSC transformation of an entire VTC dataset:

  psctrans_vtc.m

  pvd = psctrans(vtc.VTCData);

For further examples, also check out the ztrans documentation; the options work in the same way.