| Author |
 Topic  |
|
|
jimmy.dias
BIAC Alum
USA
210 Posts |
 Posted - Apr 16 2003 : 11:52:39 AM
|
Hey all,
This posting contains a summary of what the current processing daemon is doing and the options that may be configured on a per experiment ID (e.g. Example.01) basis. In order for anyone to change options for an experiment, I would need an email from the experiment PI detailing what options need to be set and the directory structure he/she would like (if different from the default). The PI can also email me (dias@biac.duke.edu) with the name of another person authorized to make these decisions.
Keep in mind that whatever the MR tech types into the PatientID field on the scanner will be used to determine the processing options for that exam. A detailed explanation of the defaults and configurable options follows:
FOR FUNCTIONALS:
Currently the following steps are applied to functional series (and any other raw files generated):
1) Raw data files (pfiles) are transferred off the scanner to the processing machine. Files are transferred 20 seconds after the pfile has been created and removed 30 minutes later.
2) Pfiles are handled as follows:
- High order autoshim files (from the 4.0 T) are archived but not transferred to Bristol and no processing is done.
- Pfiles that are reconstructable (ie those that use spiral, invspiral, and epiduke recons) are archived but NOT transferred over to Bristol. The following processing steps are done:
a) The pfile is reconstructed using either spiral, invspiral, or epiduke reconstruction techniques based on the psd name.
b) The resulting files are reformatted into volume format.
c) A BIAC XML header (BXH) describing the reconstructed data is generated, named run%{s}_%{r}.bxh and put in the directory (see http://www.biac.duke.edu/research/bxh/ for details)
d) Generate stability data and stability plots.
e) These files are transferred to the following directory on Bristol: Func/%{d}_%{e}/run%{s}_%{r} (see note below for variables available to construct file and directory names).
For example:
\\Bristol\BIAC\Group\Example.01\Data\Func\20030101_12345\run004_02
- Raw data files that are not reconstructable are archived and transferred over to Bristol (default: Pfiles/%{d}_%{e}/)*
The following options for the functional data processing stream may be set per experiment ID.
1) Toggle on/off transfer of High order autoshim files (default= off)
2) Specify the name of the BXH header. (default= run%{s}_%{r}.bxh)*
3) Toggle on/off transfer of ALL pfiles and specify directory name. (default: Pfiles/%{d}_%{e}/)*
For example:
\\Bristol\BIAC\Group\Example.01\Data\Pfiles\20030101_12345\<pfile name>
4) Toggle on/off reconstruction
5) Toggle on/off reformatting to volumes. All image data will be kept in a single file, described by an XML header.
6) Specify the directory name for reconstructed data (default: Func/%{d}_%{e}/run%{s}_%{r})*
7) Toggle on/off generation of stability data. Stability plots and text files will only be generated currently with the reformat to volume enabled. By default, these files will be generated with pfiles that are reconstructed. The name of the plot and the name of the text file may be specified.
FOR ANATOMICALS:
Currently, the following rules applies for anatomicals:
1) Anatomicals are automatically pushed per series off the scanner to the processing machine and archived.
2) Anatomicals are handled as follows:
- There are currently no processing steps for the anatomical stream.
- A BXH file is created per series and transferred with the data.
The following options for the anatomical data processing stream may be set.
1) Specify the name of the BXH header. (default= series%{s}_%{a}.bxh)*
2) Specify directory in which the anatomicals and bxh files are copied to. (default: Anat/%{d}_%{e}/series%{s}/)*
For example:
\\Bristol\BIAC\Group\Example.01\Data\Anat\20030101_12345\series003
3) Toggle on/off compressing (zip) files before transferring to Bristol. (default: off). Specify name of zip file: (default: Anat/%{d}_%{e}/series%{s}_%{r}.zip)*
*NOTE, Variables available to construct file and directory names
SAMPLE %{p}/exams/%{d}_%{e}/
%{e} exam number
%{p} patient id (eg Example.01)
%{d} date (yyyymmdd)
%{s} series number
%{r} run number (functional series only)
%{a} acquisition number (anatomical series only)
%{l} gems suite id (eg mr5c, MR6C)
Any questions/concerns pertaining to the above procedures, please post to the forums as I am sure other people might be wondering the same thing.
|
Edited by - jimmy.dias on Apr 16 2003 6:48:19 PM |
|
|
rachel.gimpel
New Member
11 Posts |
Posted - Apr 24 2003 : 10:49:53 AM
|
Jimmy,
Where do Diffusion Tensor images fall in this pipeline? I assume they follow the same pipeline as the anatomicals, and we have successfully received a few of DTI series this way. However, for our last two scans, the DTI series were not transferred to Bristol. All other anatomicals were transferred successfully. Any thoughts?
Thanks.
Rachel |
 |
|
|
jimmy.dias
BIAC Alum
USA
210 Posts |
Posted - Apr 24 2003 : 2:15:34 PM
|
The only way a series might be missing on Bristol is if 1) the scanner did not automatically transfer these particular files, or 2) if they were indeed sent by the scanner, the DICOM receiver on the processing machine did not accept them.
So I have a couple questions for you.
1) Is this a problem thats pretty consistent? ie Is there anything in particular about these DTI series that differentiate it from the ones that were successfully transferred?
2) Is it a whole series of images that are not transferred, some portion of images within a series, or both? If it is some portion of images within a series, is it a contiguous set of images that are missing or does it seem to be random? And if its a contiguous set of images within a series, is it in the front, middle, or end of the series?
We're currently using a slightly modified version of DCMTK's biacscp as the DICOM receiver. I'll post to their discussion lists as soon as we can qualify the issue. If it is the DICOM receiver thats the problem, we'll look into installing another DICOM receiver. However, if its GE's automatic transfer software thats causing the problem, we'll have to submit a bug report and see how soon they can get back to us.
|
Jimmy Dias
Software Engineer
Brain Imaging and Analysis Center
Duke University
Durham, NC 27710 |
|
|
|
rachel.gimpel
New Member
11 Posts |
Posted - Apr 25 2003 : 4:01:15 PM
|
In answer to your questions:
1) I am not sure yet if it is a consistent problem. We just recently changed our DTI sequences for the Autism.01 study. Prior to the change, we had problems with DTI transfer in which the series would transfer but not all of the slices. Since the change in sequences (unrelated to the transfer problem), we have only done 3 scans. The last 2 of those 3 did not transfer. I do not see any pattern.
We have the same DTI sequences for our FragileX.01 study, and they seem to be transferring fine so far.
2) The entire series did not transfer.
Let me know if you have other questions.
Thanks.
Rachel |
|
|
| |
Topic |
|
|
|
BIAC Processing Pipeline Options
