LONI: Laboratory of Neuro Imaging


ICBM Subject Database

The ICBM Subject Database is a web-based database infrastructure that simplifies image dataset collection, organization and dissemination. A web interface provides the means to query the data base using a combination of subject demographics and scan-related attributes. Authorized users may view representations of the data and form collections of datasets that can be downloaded or fed directly into the Pipeline environment for distributed processing and analysis.

The Database has been constructed to provide an effective means for archival and protection of collaborator collected image data. The goal of this software is to provide a convenient mechanism for searching the existence of particular image data while protecting its usage at the same time. Browse the Database


The MITRE corporation has designed, built, and deployed a full-featured ICBM DTI data management system in order to facilitate the creation of probabilistic maps of white matter. This data management system will provide a central repository for 300 identically-acquired diffusion-weighted image (DWI) datasets from five different ICBM laboratories. The system is based on the NeuroServ system, built from heavily-tested free components including Postgres, Tomcat, and middleware developed by MITRE. The MITRE system provides a means of handling the worldwide distribution of participants, the need for uniform treatment of all datasets, the need for intuitive data access and manipulation, the need for a management infrastructure appropriate to the size and importance of this dataset (ensuring its future reuse), and procedures for limiting access to the data.

Data Submission and Processing

Data submission is performed over the internet. Users enter all relevant metadata directly into a web form. The underlying software inserts them into the database. After metadata entry, the associated raw DWI file is sent to the database via a browser-initiated file transfer.

The submitted raw datafiles are processed within the database and newly derived files are linked to their precursors. The processing interface is web-based. DWI processing involves three steps: a) intra-subject registration to correct for head movement, b) inter-subject co-registration for stereotaxic alignment of all subjects, and c) tensor map computation.

At the conclusion of data collection and processing, the database will contain 300 individual tensor fields linked to their precursors and to metadata describing the original acquisition.

In This Section