SPM/Programming intro


This page is intended to provide a quick-start guide to writing your own MATLAB scripts and functions using SPM as a library. It's based on one of John's Gems, but will hopefully grow into a more complete introduction. SPM programming can mean simply writing batch scripts to automate common pipelines, writing short helper scripts or functions to accomplish useful tasks (e.g. see the other Gems, including those for SPM2 and SPM99), writing your own SPM extensions, or even modifying your local installation of SPM.

This page unashamedly fails to provide for older versions of SPM or peculiar versions of MATLAB --- it is simply too much work for the developers and keen mailing list contributors to support multiple releases. Having said that, if a function listed below exists in SPM2 or even SPM99, then there is a reasonably good chance that it will behave similarly, but you have been warned that it might not...

Using SPM from the command lineEdit


SPM has extensive developer documention in the headers of the source files. To view the documentation of a function either open the corresponding source file or call help function from the MATLAB command line. Make sure that the SPM folder is in MATLABs search path.

Reading image headers and dataEdit

  • spm_vol – header information
  • spm_read_vols – for reading entire volumes (see also: spm_vol)
  • spm_slice_vol – for arbitrary planes
  • spm_sample_vol – any voxels
  • spm_get_data (spm_sample_vol) – any voxels from multiple volumes
  • spm_bsplins (spm_bsplinc) – NB spm_slice_vol and spm_sample_vol offer polynomial or sinc interpolation; these functions provide b-spline interp as used in spm_reslice

Writing dataEdit

  • spm_write_vol (spm_vol)
  • spm_create_vol
  • spm_write_plane

Reading and writing data with the alternative NIfTI class libraryEdit

  • nifti (@nifti/Contents, @nifti/create)
  • file_array (@file_array/Contents)

Geometry, voxel-world mappingsEdit

  • spm_get_space – get the voxel-world mapping matrix (a rigid or affine transform, in homogeneous coordinates)
  • spm_imatrix – convert above matrix to parametrised form
  • spm_matrix – convert parameter vector to affine matrix
  • spm_check_orientations

Linear (rigid/affine) registration and reslicingEdit

  • spm_realign
  • spm_coreg (spm_reslice)
  • spm_affreg
  • spm_reslice – needs reference image; see following for reslicing to specified geometry
  • reorient (resize_img) – available from John's Gems 2 and 3

Preprocessing, including segmentation and non-linear normalisationEdit

  • spm_preproc (spm_config_preproc, spm_prep2sn, spm_preproc_write) – SPM5's unified segmentation and normalisation
  • spm_normalise – the old pre-SPM5 non-unified spatial normalisation
  • spm_segment – the old pre-SPM5 non-unified tissue segmentation
  • spm_smooth


  • spm_imcalc – perform arbitrary calculations on volumes (low level function)
  • spm_imcalc_ui – high level convenience wrapper for spm_imcalc


  • spm_ancova (spm_reml_ancova) – unused by SPM itself, but often useful for scripting or educational purposes

Viewing dataEdit

  • spm_check_registration (spm_image, spm_orthviews) – the ubiquitous three orthogonal views
  • slover – slices through images, overlays of thresholded or raw statistics; see also slice_overlay
  • spm_mip_ui (spm_mip) – maximum intensity projections or glass brain images


  • spm_defaults

Batch SystemEdit

  • spm_jobman

Illustrative examplesEdit

Reading and writing a volume (to replace NaNs with zeros)Edit

A simpler (but more memory-hungry) version of an old Gem. See the SPM8 version of gem for an example of plane-wise reading and writing.

fnm = spm_select(1, 'image');
[pth bnm ext] = spm_fileparts(fnm);
VI = spm_vol(fnm);
VO = VI; % copy input info for output image
VO.fname = fullfile(pth, [bnm '_zn' ext]);
img = spm_read_vols(VI);
img(isnan(img)) = 0; % use ~isfinite instead of isnan to replace +/-inf with zero

External linksEdit

See also:

Last modified on 6 May 2013, at 16:20