SCIRun User Mailing List

Text archives Help

Re: [SCIRUN-USERS] BioTensor Power App questions

Chronological Thread 
  • From: David Weinstein <>
  • To:
  • Cc:
  • Subject: Re: [SCIRUN-USERS] BioTensor Power App questions
  • Date: Sat, 16 Jul 2005 14:31:56 -0600

Hi Ashu,

We have a reader in SCIRun that can import native MATLAB files; however, it is not yet integrated into BioImage. For the moment, you will have to do this as a two-stage process: 1) convert your MATLAB data into NRRD format, 2) read that .nrrd into BioImage. I will describe how to do these below.

Regarding the dimensions of the data: it's easiest if you save your data as grid of floats with size 7 x NX x NY x NZ (where NX, NY, and NZ are the sizes of your data grid). The order of the entries along that first 7-entry axis should be: "C Dxx Dxy Dxz Dyy Dyz Dzz". C is a "confidence" value (it's used to mask out the background) -- you should use a value of 1.0 for voxels within the head volume, and a value of 0.0 for voxels outside (background). The other six entries are the upper-right triangle of the tensor matrix (since the matrix is symmetric, those six are sufficient to define the tensor). If you need assistance in any of this data formatting, I would be happy show you how to insert additional "Unu" formatting modules into the net I describe below.

To accomplish 1) --
    - launch SCIRun
    - create a MatlabInterface::DataIO::MatlabNrrdsReader module
- on the MatlabNrrdsReader UI, point the top text entry field at your .mat file and press the Execute button
- create a Teem::UnuAtoM::UnuAxInfo modue, and pass the first output from MatlatNrrdsReader into it
- on the UnuAxInfo UI, pull down the "Kind" menu and select "nrrdKind3DMaskedSymMatrix"
- create a Teem::DataIO::NrrdWriter module, and pass the output from UnuAxInfo into it
- on the NrrdWriter UI, type in the name for the output file (any name will do, e.g. /tmp/tensors.nrrd), and press the Execute button
    - close SCIRun

Now your data is in a format that BioTensor supports. When you launch BioTensor, select the "Tensor Volume" radio button on the Load Data tab, and then on the "Generic" tab (just below) use the DWI Volume's "Browse" button to load in the NRRD that you created with the SCIRun network (press the "Set" button in the browse UI once you've located the data). Finally, press the blue Execute button (at the bottom of the Execute tab) and after a few seconds you should see your data in the 3D View Window.

Please let me know if you run into any problems with this.


P.S. To answer your second question: the demo-DWI.nrrd file is a collection of diffusion-weighted images (the raw data from which diffusion tensors can be built). The first Processing stages of BioTensor allow users to load in raw DWI data and to convert them into approximating tensors (the DTI volume). For more information, please see our BioTensor tutorial -- User/Tutorials/BioTensor/BioTensor.html

On Jul 15, 2005, at 11:59 AM, Ashutosh Chaturvedi wrote:


I was wondering if any of you have used the BioTensor Power App? I had
a couple of questions regarding its use for fiber tracking within DTI

1. I have a raw DTI dataset in a 3D matrix form which contains a 3x3
tensor at each element within the matrix. I can also change the tensors
at each element to be 1x9 instead of 3x3 using MATLAB if necessary.  I
was wondering how I can import that into a tensor volume or nrrd dataset
that can be read by BioTensor?

2. The example DTI data demo-DWI.nrrd that's given within the SCIRun
datasets brain-dt has four dimensions.  The first dimension is 12,
whereas the other three dimensions correspond to the size of the
dataset.  What does that first dimension mean physically?  Should my
dataset have the same size first dimension?

Thanks in advance for your help!

--Ashu Chaturvedi
====================================================================== =====
== The SCIRun Users mailing list: send email to ==
== for more details. ==
== Please acknowledge use of SCIRun in your papers and reports: ==
== see biopse_citation.bib ==
====================================================================== =====

== The SCIRun Users mailing list: send email to   ==
== for more details.                                                     ==
== Please acknowledge use of SCIRun in your papers and reports:          ==
==   see         ==

Archive powered by MHonArc 2.6.16.

Top of page