A Modular Image Transfer, Archive, Database System
for Research and Clinical Image Distribution
R.L. Levin1, G.S. Sobering1, M. Douglas1, S. Misra2, J. Conner3, J.A. Frank4
1-Multi-Modality Radiological Image Processing Section, NIH, Bethesda, MD
2-Sensor Systems, Inc., Sterling, VA
3-DeJarnette Research Systems, Inc., Towson, MD
4-Laboratory of Diagnostic Radiology Research, NIH
Submitted to RSNA 1995 Annual Meeting
Purpose:
This project provides a mechanism for researchers and clinicians to access
images collected at the NIH. The system is primarily designed to
automatically receive DICOM images from radiological scanners, but can also
accept other formats. Images are archived in their native format. A simple
data-base provides secure access to the image archive.
Material and Methods:
Images from DICOM-capable scanners are automatically sent to the
"catcher" program (DeJarnette) running on a small UNIX workstation, which
has adequate local storage for one day's images. The "catcher" extracts 30
header fields for the data-base, and requests the "committer" program
(Sensor) to add the images into the archive. The "committer" copies the
images into the Andrew File System (Transarc) archive and creates a data-base
entry. Access to the data-base is controlled by the Kerberos authentication
system.
Results:
Typically, less than 45 minutes after a scan is completed, the images have
moved through the "catcher" and "committer" and are available in the
archive.
Conclusion:
Open systems and standards (ex. UNIX, TCP/IP, DICOM) make it possible
to construct an non-proprietary image acquisition and archive system. This
system is currently archiving ca. 4 gigabytes of data every day.
Financial Disclosure:
S. Misra in an employee of Sensor Sysytems, Inc. and J. Conner is an
employee of DeJarnette Research Systems, Inc.