Hi John,
If you don't have the voxel dimensions set correctly then I would expect
you to have problems with BET like you describe.
There are two things to do about this.
Firstly, see if there is an easy way to fix the ImageJ conversion.
To do this, create your own analyze header using avwcreatehd.
Then compare your header with the one from ImageJ (you can list the
fields with avwhd) and if there is a difference in some of the critical
fields like dim1 to dim4, pixdim1 to pixdim3, sizeof_hdr or datatype then
this is the problem and you can easily write a script using avwval and
avwcreatehd to fix the problem.
The second alternative is to just use avwcreatehd every time as part
of your conversion process. To do this you would need to get the
appropriate values (like voxel dimensions) from somewhere. Once you
have the header file all the .img file must contain is a
straightforward binary list of intensity values, voxel by voxel,
incrementing x first, then y, then z. So, if you can get something to
save the image data like this then that's all you'll need for a
successful conversion.
All the best,
Mark
John Grinstead wrote:
> I was having problems using BET (it was drawing the contours much too
> small) when I used images that I had converted from Dicom to Analyze format
> with ImageJ (using the Analyze Writer plugin from
> http://rsb.info.nih.gov/ij/plugins/analyze.html to write out a stack of
> T2-wtd anatomical images). When I converted them with MRIcro then BET
> worked fine. Anyone had this problem and know how to fix it within ImageJ?
> Perhaps ImageJ throws out some header information (such as field of view)
> that BET requires?
>
> This is really only an issue for me because I run FSL on a Mac, and MRIcro
> is for windows only. If someone knows a reliable way to convert these files
> with some other simple to use Mac friendly software that would be great.
>
> Thanks,
> John
|