Methods
The storage workflow of the Data Grid is shown as Figure 1:
-
Field Sites (1-n) send images to Grid Access Point (GAP) A of the Site A in the Data Grid Radiology Core through DICOM transmission. GAPs B and C can also be used to receive images from these Field Sites if GAP A is down, therefore avoid single point of failure.
-
The GAP A stores the metadata of the images in a Metadata Database. The Metadata Database must support dynamic creation of new tables or alteration of existing tables. Thus, new trials can be added to the database anytime without affecting the existing trials.
-
The images are archived to the image Repository A, which can be a RAID (Redundant Array of Inexpensive Disks) or SAN (Storage Area Network). Each image repository in the Grid is separated into two partitions. Partition 1 (P1) is used to store the images received by this site. Partition 2 (P2) is used to store the backup of images from other sites.
-
The GAP A replicates one copy of the images to the P2 of the Repository B. Two copies of an image will always be available within the Data Grid in order to avoid single point of failure. The images can also be replicated to the P2 of the Repository C. Where to replicate the images is defined by a replication policy table consisting of various criteria, such as storage capacity or trial requirement. The replica location information is stored in a Replica Database.
Figure 1 Storage workflow of image based clinical trials using Data Grid. A radiology core containing three sites (A, B, and C) can store, backup, and share images in the Data Grid. Repositories A, B, and C are physically separated into P1 and P2. The images of Site A are archived in P1 of the Repository A and backed up in P2 of the Repository B or C. The Metadata Database and Replica Database are shared by the three sites.
|