If you can't wait for Rick's improved script, you can try 3dretroicor. I haven't worked on it in several years, but it's still there. Your mileage may vary. Subtract the output from the input dataset to see what it did to each voxel timeseries.
If you take the (optional) phase outputs from 3dretroicor I believe you can calculate RETROICOR regressors yourself by simply computing the sine and cosine. The annoying part is downsampling the regressors to the times of slice aquisitions. Maybe I could add that to 3dretroicor sometime.
