File formats

The formats Larmor reads and writes, with the Python functions for each. The load_ and save_ functions of larmor.io, and the lar commands, choose the format from the end of the file name; the read_ and write_ functions of larmor.io.formats read or write one format.

Images

Format	File names	Read	Write	Python
NIfTI-1 and NIfTI-2	.nii, .nii.gz	yes	yes (NIfTI-2 only when a dimension needs it)	load_volume(), save_volume(), read_nifti(), write_nifti()
MRtrix image (MIF)	.mif, .mif.gz, .mih with its .dat	yes	yes	load_volume(), save_volume(), read_mif(), write_mif()

A MIF header can hold the gradient table (dw_scheme), which load_dwi() reads with the image and save_volume() writes into it.

Gradient tables

Format	File names	Read	Write	Python
FSL bvals and bvecs	any, often .bval and .bvec	yes	yes	read_bvals_bvecs(), write_bvals_bvecs()
MRtrix gradient table	any, often .b	yes	yes	read_mrtrix_gradients(), write_mrtrix_gradients()
b-matrix, one row per volume	any	yes	yes	read_b_matrix(), write_b_matrix()

Tractograms

Format	File names	Read	Write	Python
MRtrix tracks (TCK)	.tck, .tck.gz	yes	yes	load_tractogram(), save_tractogram(), read_tck(), write_tck()
TrackVis (TRK)	.trk, .trk.gz	yes	yes (version 2)	load_tractogram(), save_tractogram(), read_trk(), write_trk()
Legacy VTK polydata	.vtk, .fib, each also .gz	yes	yes	load_tractogram(), save_tractogram(), read_vtk(), write_vtk()
MRIToolkit tractography	.mat	yes	no	load_tractogram(), read_mritoolkit_tracts()

Large uncompressed TCK and TRK files can be read and written batch by batch, with iter_tractogram() and open_tractogram_writer().

Values along streamlines

Format	File names	Read	Write	Python
MRtrix track scalar file (TSF): one value per point	.tsf	yes	yes	read_tsf(), write_tsf()
Streamline weights, as tcksift2 writes them: one value per streamline	any text file	yes	yes	read_streamline_weights(), write_streamline_weights()

Surfaces

Format	File names	Read	Write	Python
GIfTI surface	.gii, such as .surf.gii	yes	yes	load_surface(), save_surface(), read_gifti_surface(), write_gifti_surface()
GIfTI values per vertex	.shape.gii, .func.gii, .label.gii	yes	yes (.shape.gii, .func.gii)	read_gifti_values(), write_gifti_values()
FreeSurfer surface	no extension, such as lh.white, lh.pial	yes	yes	load_surface(), save_surface(), read_freesurfer_surface(), write_freesurfer_surface()
FreeSurfer values per vertex (“new curv”)	such as lh.curv, lh.thickness	yes	yes	read_freesurfer_values(), write_freesurfer_values()
FreeSurfer label	.label, such as lh.cortex.label	yes	no	read_freesurfer_label()

Registrations

Transforms are read, not written; load_transform() tells the format from the file where it can.

Format	File names	Read	Write	Python
ANTs (ITK) affine	binary .mat (such as 0GenericAffine.mat), text .txt, .tfm	yes	no	read_ants_transform()
ANTs displacement field	NIfTI, such as 1Warp.nii.gz	yes	no	read_ants_transform()
FSL FLIRT matrix	text, such as .mat from -omat	yes	no	read_flirt_transform()
FSL warp field	NIfTI (fnirt --fout, convertwarp, invwarp)	yes	no	read_fsl_warp()
MRtrix3 linear transform	text (mrregister, transformconvert)	yes	no	read_mrtrix_transform()
MRtrix3 deformation or displacement field	.mif or NIfTI	yes	no	read_mrtrix_warp()

MATLAB and MRIToolkit

Format	File names	Read	Write	Python
MATLAB file, versions 4, 5 to 7 and 7.3	.mat	yes	no	read_matlab()
MRIToolkit DTI or DKI fit	.mat	yes	no	read_mritoolkit_fit(), read_mritoolkit_map()
MRIToolkit image, its values as stored	.mat	yes	no	read_mritoolkit_volume()
MRIToolkit FODs	.mat	yes	no	read_mritoolkit_fod(), read_mritoolkit_sh()
MRIToolkit peaks	.mat	yes	no	read_mritoolkit_peaks()

Tables and other text files

Format	File names	Read	Write	Python
Lookup table, FreeSurfer or MRtrix layout	text, such as FreeSurferColorLUT.txt	yes	no	read_lookup_table()
Connectome matrix with region names	.csv, or text separated by spaces	no	yes	write_connectome()
Regions of each streamline’s endpoints, as tck2connectome writes them	text	no	yes	write_node_assignments()
Seeds, one row per seed	text	yes	yes	read_seeds(), write_seeds()
Tract profile, one row per point	.csv	no	yes	TractProfile.write_csv
BrainViewer session	.lar.json	yes	yes	Viewer.load_session, Viewer.save_session