XNAT Project IDs

Introduction

If you manage a research project that employs MR imaging, you can benefit greatly by integrating your scan sessions into XNAT, the Extensible Neuroimaging Archive Toolkit (http://xnat.org for more details). We have an XNAT instance here at MIT, which can be accessed at http://xnat.mit.edu . You can request an account from the home page - though please note, only people who are actively affiliated with the MIT McGovern Institute will have their requested account activated.

Once you are in the system, you can begin tagging your MR sessions (details below) so that they will get automatically uploaded to the Mindhive compute cluster. Although you should always, always be responsible for ensuring that your sessions get transferred properly - this responsibility is solely the researcher's -it can be a big convenience to have your images automatically go to the right place on the cluster file system.

Additionally, having your images in XNAT allows you to catalog your projects, search for particular session attributes (e.g., query the database for all sessions of right-handed males under the age of 30) and view QA mosaics of sessions. Dicom header information for images also gets transferred in, so you can look at things like the slice prescription and other information in a user-friendly interface.

How to Tag Your Sessions

It's quite simple. You need to agree in advance with everyone in your project that you will consistently use a Project ID every time you scan a subject. The ID should be something short and memorable that accurately characterizes the project. For example, GAB_SAD could represent the SAD project within the Gabrieli lab. KAN_EK could represent the Ellison Kids project in the Kanwisher lab.

Once you have settled on your permanent ID for your project, you should communicate it to Mark Pearrow, who will set up a project instance for you on XNAT. Then, for all future scan sessions, you should include a special tag in the "Additional Information" section under "Patient Info" on the patient card dialog at the scanner console. Figure 1 depicts the location of this field in the Syngo (scanner console) screen.

Siemens Syngo scanner console.
Figure 1: The Syngo console, showing the location of "Additional Info" under "Patient".

The format of the tag is as follows:

Project:<PROJ_ID>

For example:

Project:GAB_SAD

There must be no spaces in this string. When the session is sent from the scanner to Sigma, it will also get automatically pushed to the Mindhive cluster. If you've added the tag for your project, as shown above, the session will get automatically moved into a directory for your project.

The DICOMS that get moved in this fashion show up at /mindhive/DICOM/archive/<PROJ_ID>. Scans that do not get tagged end up in a sort of XNAT purgatory, called the "prearchive", and these scans can be found at /mindhive/DICOMS/pre_archive/.

Great, Now What?

Once your DICOMS are on the cluster and in XNAT, you can access them two different ways. One is by viewing them in the XNAT web interface, which gives you access to a nice GUI and lots of tools for inspecting them and doing queries on them, as well as adding information to each subject, session, and project. In the near future, you will also be able to send a session or multiple sessions to a preprocessing and analysis pipeline.

You can also choose to access the files via the Mindhive filesystem. Note that /mindhive/DICOMS is a read-only filesystem. You cannot alter the contents in any fashion by direct manipulation on the cluster. This is by design; we want to be able to keep pristine originals of data and constrain them to a particular, well-ordered file system layout.

You can create symbolic links to the DICOM directories in your analysis directories, however. This way each person who is doing analysis can just point to the canonical DICOM directory, rather than making numerous copies of the same data.