|
Imaging researchers are faced with complex challenges of integrating huge volumes of data obtained from a wide range of distributed and uncoordinated systems, and sharing images and data with multidisciplinary teams within and across institutions. Images must be aligned, registered, and integrated with data from other sources, then analyzed, manipulated, and rendered using in-house software or scripts written for environments such as SPSS, Analyze or MATLAB. Files are typically distributed over a range of machines and stored in ad-hoc directories. This may be adequate for an individual researcher or a small tightly coordinated team, but when research requires cooperation between autonomous groups, bottlenecks can occur if specific personnel are unavailable to perform critical functions such as retrieving a file or generating a data set.
We have developed web-based software called BioSCRIBE that allows neuroimaging investigators to more easily acquire and manage multimedia research data, and to selectively share data and images with interdisciplinary collaborators. The software enables investigators to securely and selectively link data and images for easy access by demographics, physiologic properties, and experiment attributes. BioSCRIBE is being developed using a formative evaluation approach, with continuous feedback from researchers at the new Integrated Brain Imaging Center at the University of Washington, a multidisciplinary neuroscience imaging laboratory.
Through secure web-based interfaces, BioSCRIBE allows geographically dispersed imaging researchers to create and maintain centralized repositories of multi-modal clinical images and their associated experiment data. Any type of file or exam can be imported into the repository, assigned metadata, and retrieved on demand through the web interface. Files can be sent via DICOM, uploaded through web browsers, or transferred to a secure FTP directory. Once in the repository, files can be tagged with keywords, grouped into web folders, and organized in structured hierarchies.
When a new study arrives in the system, it is automatically routed to the appropriate project Inbox based on metadata such as DICOM header information. Email alerts keep users informed of events, such as the arrival of a series that matches a subject registered to a specific project. Project members are prompted to enter additional metadata, such as study parameters and analysis type. Images and studies can be categorized by anatomy, pathology, modality, and experiment parameters. Any object or file in the system can be retrieved via full text search, or through advanced queries that allow users to compose constraints on combinations of attributes. To support user-friendly browsing and searching, BioSCRIBE generates a navigation trail at the top of each page, providing a visual context for traversing through hierarchical data.
BioSCRIBE is HIPAA compliant through use of strict access control, encryption, and audit trails, providing a full log of every operation performed by every user. Access to images, studies, and experiment data are controlled through a digital asset management interface that allows users to specify who can see or edit objects at a fine level of granularity. For each project, users may specify core project members and collaborators, and control access based on each user's relationship to the data. Patient identifiers and accession numbers can be included or withheld from each view on a "need to know" basis.
BioSCRIBE is in the early stages of deployment at the Integrated Brain Imaging Center, where it will support a wide range of imaging projects. Future development will focus on managing imaging pipelines through a scripting interface to support custom workflow. Once the system is ready for widespread distribution, it will be an important tool in reducing barriers to interdisciplinary and cross-facility imaging research.
This research is funded by NIMH SBIR R44-MH070166. The University of Washington Integrated Brain Imaging Center is funded by a grant from the Murdock Charitable Trust. |
