M. Vink

Design Magic (d_magic) calculates the multicolinearity indices for all factors of a design matrix generated in SPM (version 2/5/8). In addition, it calculates the optimal cut-off for the high-pass filter commonly used in fMRI research to remove low drifts in the signal.

The idea behind the program is that is allows users to evaluate their design in terms of uniqueness of the factors. In this way, one can make an optimal design prior to scanning, in terms of minimising overlap between factors. Another issue important for obtaining an optimal design is the amount of variance for each factor. This is not calculated, but will be incorporated in the next version of the program.

After selecting an SPM.mat file, the program will calculate the R^2 value for each factor. This value indicates how much variance is shared with the other factors in the design. An R^2 value of .33 means that 33 percent of the variance of that factor is also explained by (any linear combination of) the other factors. Generally, it is a good thing to keep the R^2 for each factor low so that each factor explains as much unique variance as possible.

These R^2 values are also used to evaluate the impact of the high-pass filter, which is made up of cosine function. The program will let you test your own high-pass filter as well as calculate the optimal cut-off for the filter.

Download