AtlasStateExchange_2007 < MouseBIRN

See The original page where exchange of Atlas State was started

Overview:

Work resumed at the end of 2006 with a focus on setting up an Atlas Server that can be used in conjunction with the API for atlas exchange. This outlines the step by step plans of the group.

Synopsis of intitial work plans:

Meeting 1/14/07 between Willy, and UCLA:

Immediate:
- Willy will give Allan the origin and define the coordinates of SA in relation to the mouse looking at you
- Allan will generate the transformations between the atlases and give Willy a transformation for a single Paxinos slice on Tuesday
- We will meet again next Thursday at 10:30 AM
Preparation:
- Willy will look into a DB where the transformations can be stored and Allan will have access to update them (the plan is to be able to access these transformations through the server web-services)
- Willy will give Steve the source code for testing access to the server
- Steve and Heng will begin testing the atlas web-services API sent by Willy and Willy will send basic documentation and a test application for accessing this service

Synopsis of follow up meeting 1/18/07:

Willy, Drexel, and UCLA:

There was still some confusion about the coordinates used by the three systems, so to clarify, we went through them all together:
- Confirmed that all three systems have the brain situated in the coordinate frame in pretty much the same manner (NT Y coordinate may be 5 at Bregma, but it still increases going down into the brain)
- It’s a little unclear if the coronal planes used in SmartAtlas^? are with the animal facing towards us or away from us, but it probably shouldn’t make too much of a difference for how we’ll likely handle the interactions
- When we pass values of X, we’ll probably stick to absolute values, as most people will want data from both the left and right sides of the brain
Next steps:
- Allan will 10 points in the brain in his atlas and will send the XYZ coordinates as well as the names of the structures that are within those coordinates to Willy, Yoni, and Bill
- They will verify that this aligns to their atlases
- Allan will send a single transformation to Willy by tomorrow (probably at Bregma 0)
- We will use this test transformation for testing purposes as outlined in the previous e-mails.
- Touch base at next week’s regular programmers VTC

Point	X	Y	Z	UCLA	UCSD	Drexel
1	2.00	4.00	0.02	Cpu
2	1.79	4.93	0.02	VP	VP
3	0.45	2.62	-3.80	PAG	PAG
4	0.46	5.22	-3.80	Pn	Pn
5	-2.42	3.94	-3.80	PoDG	PoDG
6	0.89	1.74	3.57	EPIA	AOL
7	-0.69	3.58	3.57	GrO	GrO^?
8	-3.28	3.92	-5.88	PFl	PFl
9	0.00	4.00	-5.88	4V	4V
10	1.22	5.61	-5.88	7	7N
11	2.00	4.00	2.00	LO	LO

2D transformations from Paxinos 2001 to MRI:

Below are the transformations in Montreal Neurological Institute XFM format. The XFM files are text files that can be read in any text editor. Though the transformations are in 3 dimensions, I have edited them to two. I was not able to get the full affine transformation nor the thin plate splines to work in 2D. This transformations are in world coordinates (not voxel coordinates), so in order to use them, you will need the step sizes and starts contained in the mnc file header. If you can't read these mnc files, we've put the relevant information from them below the file download link.

Since the transformation is in world coordinates and works off the Paxinos coordinate system, I expect that simply applying the transformation matrix to Paxinos coordinates should yield the appropriate location in the 2D MRI image. We will have to repeat this for every plate; tedious, but doable.

I used the MNI MINC Tools to align the images and manipulate the transformations. They tools can be downloaded at: http://www.bic.mni.mcgill.ca/software/distribution/

The transformations were produced as follows:

Created a 3D volume from the Paxinos mouse brain atlas plates by scanning in the delineations and spacing them with blank planes.
Aligned MRI volume to Paxinos volume using Register to calculate a full-affine transformation (12 parameter) by picking fiducial points and Mincresample to apply the transformation and resample into the same voxel space as the Paxinos volume.
Separated both volumes in to their component 2D image with Minctoimage and selected Paxinos image closest to bregma (0.02) and corresponding MRI image.
Aligned MRI image to Paxinos image using Register to calculate a rigid body (6 parameter) and a 9 parameter transformation by picking fiducial points and Mincresample to apply the transformation (already in the same voxel space as the Paxinos image).
Manually edited the transformations to remove Z dimension transformations with Xfm2param and Param2xfm.
Inverted transformations so that they convert coordinates from Paxinos atlas to MRI, rather than the other way around, with Xfminvert.

PaxinosBregma.mnc: Paxinos Atlas image in MNC format
X dimension 606
Y dimension 500
Z dimension 1
X step 0.01818
Y step 0.01818
Z step 0.01818
X start -5.50854
Y start -1.21806
Z start 0.02

Pax2MRI6.xfm: Rigid body transformation (6 parameters, 3 translations and 3 rotations) aligning Paxinos atlas to MRI

Pax2MRI9.xfm: 9 parameter transformation (3 translations, 3 rotations, and 3 scales) aligning Paxinos atlas to MRI

Webservice for transformations:

Created by Willy Wong

Database-driven and its schema can be re-used for the different kind of transformation (4 x 4 matrix)
Go here for the WSDL
Go here for testing
Transformation data at bregma = 0.02 as of 2/09/07
Atlas Interoperability API

Paxinos plates transformations:

Created by Allan MacKenzie-Graham

Transformations for Paxinos plates 02-93 - 01 and 94 forthcoming.
Transformations 95-100 may involve starting from scratch. Paxinos atlas 3D volume was created with the first edition of the atlas.
Transformations 02-93

Attachment	Action	Size	Date	Who	Comment
PaxinosBregma.mnc	manage	297.7 K	01 Feb 2007 - 20:24	AllanMackenzieGraham	Paxinos Atlas image in MNC format
Pax2MRI6.xfm	manage	0.2 K	01 Feb 2007 - 23:27	AllanMackenzieGraham	Rigid body transformation (6 parameters, 3 translations and 3 rotations) aligning Paxinos atlas to MRI
Pax2MRI9.xfm	manage	0.2 K	01 Feb 2007 - 23:28	AllanMackenzieGraham	9 parameter transformation (3 translations, 3 rotations, and 3 scales) aligning Paxinos atlas to MRI
AtlasInteroperabilityAPI.ppt	manage	115.5 K	27 Feb 2007 - 22:50	QueenieNg	Atlas Interoperability API
XFM.zip	manage	21.4 K	07 May 2007 - 20:00	AllanMackenzieGraham	Transformations for Paxinos plates 2-93