Diffusion MRI

Diffusion-weighted magnetic resonance imaging (DW-MRI) is a radiological technique that measures water diffusion in living tissue in order to infer microstructural details of fibrous textured tissues like brain white matter. The information that comes from a DW-MRI scan sequence generally requires extensive processing before it is useful for brain scientists --- see the article on diffusion MRI techniques for more details. Our research group collaborates with scientists around the world to help make medical discoveries while developing novel techniques to process, quantify, and visualize diffusion MRI data.

You may also want to check the diffusion MRI glossary page to learn some jargon, or the brain anatomy page.

Neuroscience Collaborations

The following is a table of all the projects on which we're collaborating, the data we've received from our collaborators, information about the data, and our progress in processing them.

Entries in the "Dataset Date Received" column may include a "[more]" link. This drops down more detail, including the path to the data and additional notes. Relative paths start at the directory /data/graphics/mri/.

In the "Status" column, the rule of thumb is to only mark a particular stage completed if it's completed for all scans (usually corresponding to patients) in the dataset. Sometimes a step will be marked complete if it's known to work for all scans but has not been completed on a minority of them, but this practice is discouraged.

In the "Resolution" and "Voxel Size" columns, dimensions are ordered XxYxZ.

In the "Orientation" columns, "L" stands for "patient's left". "R" stands for "patient's right". "A" stands for "anterior" (front), "P" for "posterior" (back), "I" for "inferior" (bottom), and "S" for "superior" (top). See the wikipedia page on anatomical coordinate systems for more information. The default orientation in the table below is "RL", "AP", and "IS"; if they are in the opposite order it indicates a direction flip.

Note that this "default orientation" is non-standard; NIfTI, Talairach, and MNI coordinate systems all specify that X increases L-to-R; Y, P-to-A; and Z, I-to-S. The default orientation is still a right-handed coordinate system (thank goodness), but it's a little wonky and may cause problems with certain software. Check out the NIfTI FAQ on orientation and Graham Wideman's helpful tutorial for more information. There's the chance that all of the orientation tags in this table have systematic errors, so a wholesale review of them might be worthwhile at some point. For the time being, take them with a grain of salt and double-check if it's important.

Collaboration Table

Project	Location	Collaborators	Dataset Date Received	Status								Scanner		Format	# Scans	Resolution (voxels)	# DWIs	Voxel Size (mm)	DWI Orientation						bvec Orientation
				C	U	D	F	M	F	T	B	F	B						-X	+X	-Y	+Y	-Z	+Z	-X	+X	-Y	+Y	-Z	+Z
Cognitive Aging	St. Louis	Rob Paul (UMSL), Tom Conturo (WUSTL)	2008-07-20 st_louis/pilot_data.2008.07.20	Y	Y	Y	Y	Y	Y	Y	Y	?	?	Analyze .img + Interfile .ifh header	1	83x110x64	36	2.0x2.0x2.0	R	L	P	A	I	S	R	L	A	P	I	S
			2010-07-09 st_louis/aging_brains.2010.07.09	Y	N	N	N	N	N	N	N	?	?	Analyze .hdr/.img + Interfile .ifh header (supplementary)	81	83x110x64	64 for d1 scans; 72 for d2 scans	2.0x2.0x2.0	R	L	P	A	I	S	R	L	A	P	I	S
Fibers at Risk	Colorado	Jack Simon	2008-08-28 Longitudinal data collected a long while ago, mailed to us on DVD.	Y	Y	N	N	N	N	N	N	1.5T	?	?	?	?	?	?	?						?
	Portland	Jack Simon (VA), Bill Rooney (OHSU AIRC)										7T	GE
HIV	Memorial Hospital	Steve Correia, Ron Cohen	?	Y	Y	Y	N	N	N	N	N	1.5T	Siemens	DICOM	?	256x256x90	?	0.8477x0.8477x1.667	?						?
	Brown Life Sci		2007-11-30 brown3t/302CK.2007.11.30	Y	Y	Y	Y	Y	N	N	N	3T	Siemens	DICOM	1	256x256x180	74	.8828x.8828x.7	?						?
			2008-07-21 brown3t/cohen_hiv_study.2007.02.07	Y	Y	Y	Y	Y	Y	Y	Y			NIfTI	13	128x128x70	74	1.76562x1.76562x1.8	?						?
			2008-10-09 brown3t/cohen_hiv_study_registered.2007.02.07	Y	Y	Y	Y	Y	Y	Y	Y			NIfTI	8	128x128x70	74	1.76562x1.76562x1.8	R	L	A	P	I	S	L	R	A	P	I	S
Kids	Hasbro / Coro	Daniel Dickstein	2010-01-13 hasbro/test_data.2010.01.13	Y	Y	Y	Y	N	N	N	N	?	?	NIfTI	4	128x128x70	65	1.765625x1.765625x1.8	R	L	A	P	I	S	R	L	A	P	I	S
Macaque	Brown	John Donoghue, Kaivon Paroo	2010-02-09 brown3t/donoghue_macaque.2010.01.21	Y	Y	Y	Y	N	N	N	N	3T	Siemens	DICOM	?	?	?	?	?						?
Macaque Histology	Caltech	Jason Kaufman	2008-05-21 caltech/macaque.2008.05.21	Y	Y	Y	Y	Y	Y	Y	Y	?	Bruker	Analyze	1	98x110x130	36	0.6x0.6x0.6	?						?
SA HIV	Cape Town	Rob Paul (PI), Bruce Spottiswoode, Paul Carey, John Joska	2007-10-22 south_africa/normal.2007.10.22	Y	Y	Y	Y	Y	Y	Y	Y	?	Siemens	DICOM Mosaic	1	127x128x70	40	1.71875x1.71875x1.7	?						?
			2008-05-09 south_africa/normal.2008.05.06.average1 south_africa/normal.2008.05.06.average2 south_africa/hiv.2008.05.09	Y	Y	Y	Y	N	N	N	N	?	Siemens	DICOM Mosaic	3	127x127x75	66	1.71875x1.71875x1.7	?						?
			2008-07-24 south_africa/hiv_positive_apathy_negative_15patients.2008.07.24	Y	Y	Y	Y	Y	Y	Y	Y	?	Siemens	4-D NIfTI	15	128x128x75	65 For each patient, there are 5 acquisitions in 13 directions each, all registered to each other. Each acquisition is a separate 4-D NIfTI file.	1.875x1.875x1.9	R	L	A	P	I	S	R	L	A	P	I	S
			2009-11-27 south_africa/rp_study.2009.11.27	N	N	N	N	N	N	N	N	?	?	DICOM Mosaic	30	128x128x70	33	1.71875x1.71875x1.7	?						?
Tumors	Edinburgh	Mark Bastin	2007-11-06 edinburgh/tumor.2007.11.06	Y	Y	Y	Y	N	N	N	N	1.5T	GE	DICOM tarball	1	128x128x72	71	2.0x2.0x2.0	?						?
			2008-07-24 edinburgh/tumor_pilot.2008.07.24	Y	Y	Y	Y	Y	~	~	~	1.5T	GE	DICOM tarball	29 1 scan each of 5 tumor patients, plus 3 scans each of 8 healthy volunteers.	128x128x72	71	2.0x2.0x2.0	R	L	A	P	I	S	R	L	P	A	I	S
			2009-11-23 edinburgh/tumor_fullstudy.2009.11.23	Y	Y	Y	Y	~	~	~	~	1.5T	GE	DICOM tarball	17	128x128x72	71	2.0x2.0x2.0	R	L	A	P	I	S	R	L	P	A	I	S
			2009-12-21 edinburgh/normals.2009.12.21	Y	Y	Y	Y	N	N	N	N	1.5T	GE	DICOM tarball	36	128x128x72	71	2.0x2.0x2.0	R	L	A	P	I	S	R	L	P	A	I	S

Brown Protocol Compatibility Study

The 3T scanner at Brown has changed its configuration several times, necessitating a study to determine whether data from these different configurations are compatible. Please see the page /Brown Protocol Compatibility Study for details.

Software

The principal pieces of software we use for research now are:

• Diffusion Processing Pipeline --- a series of small programs that turns data we get from collaborators into tensor models, scalar measures, tractograms, etc.
• Brainapp --- a GUI program for viewing tractograms and computing scalar statistics on them.
• $G/bin/mriatlas --- generates a PDF of slices of MRIimage files.
• Diffusion MRI Matlab Scripts --- Matlab scripts to read, write, view, and manipulate MRI data (especially in MRIimage format)

You may also be interested in the exhaustive list of diffusion MRI software we've developed in-house, or that of 3rd-party software.

File Formats

Format	Extension
DICOM	.ima; .dcm
Analyze	.hdr / .img
NIfTI	.nii; .nii.gz; .hdr / .img
Interfile	.ifh / .img; .i33 / .h33
MRIimage	none

HOWTOs

• Set up your environment for diffusion MRI processing
• Run the diffusion processing pipeline at low resolution
• Diagnose tensor orientation problems
• Compute apparent diffusion coefficient

See Also

• Introductory Reading List --- a minimal guide of good references for understanding the background of this project
• Diffusion Processing Pipeline --- a series of small programs that turns data we get from collaborators into tensor models, scalar measures, tractograms, etc.
• Brainapp --- a GUI program for viewing tractograms and computing scalar statistics on them.
• Diffusion MRI Matlab Scripts --- Matlab scripts to read, write, view, and manipulate MRI data (especially in MRIimage format)
• Diffusion MRI techniques --- technical information about the diffusion MRI process
• Brain anatomy
• Diffusion MRI glossary
• Work With Histology Data
• Diffusion MRI Design Notes
• Brain Networks
• Neuroimaging Systems