ztrans

# Differences

This shows you the differences between two versions of the page.

 — ztrans [2010/06/11 02:04] (current)jochen created 2010/06/11 02:04 jochen created 2010/06/11 02:04 jochen created Line 1: Line 1: + ====== ztrans ====== + ===== Motivation ===== + Under certain circumstances (e.g. data that is arbitrarily scaled and comes from several sources has to be compared), one way to normalize (re-scale) the data is to apply z-transformation (setting the mean of values in the desired dimension to 0 and the standard deviation to 1). + + ===== Function reference ('help ztrans'​) ===== + <​file> ​ ztrans ​ - perform z-transformation on time course + + FORMAT: ​      [ztc, zf, zsh] = ztrans(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: + + ztc         ​z-transformed time course + zf          z-transformation factor + zsh         ​z-transformation shift + + See also psctrans​ + + ===== Usage examples ===== + * z-transforming a single time-course (e.g. of a region or voxel):<​code matlab>​ztc = ztrans(tc);​ + * z-transformation of an entire VTC (ensuring that dimension is 1):​vtc.VTCData = ztrans(vtc.VTCData,​ 1);​ + * z-transformation of 4D-NII VoxelData (ensuring that dimension is 4):​nii.VoxelData = ztrans(nii.VoxelData,​ 4);​ + * also obtaining the information about the scaling factor and shift:<​code matlab>​[ztc,​ zfactor, zshift] = ztrans(tc);​ + * only using values where no experimental condition in an SDM is active:<​code matlab>% find timepoints to use + tp = all(abs(sdm.SDMMatrix(:,​ 1:​sdm.FirstConfoundPredictor-1)) < 0.1, 2); + + % apply z-transformation with those as normalization basis + vtc.VTCData = ztrans(vtc.VTCData,​ 1, tp); 