We are introducing a new database-system (DBS) currently under development at the Department of Medical Computer Sciences designed to meet the demands of image- processing researchers.
Dealing with digital images becomes more and more a part of the daily work of physicians. Specially when measuring, viewing and fusing 3-dimensional data, the advantage of digital image processing and understanding methods becomes obvious. However these methods are not integrated in commercially available database systems. Therefore, we developed a DBS which is able to perform and store image processing algorithms as well as a variety of medical images..
There are two main goals in this project. First is to establish an image exchange platform for researchers doing interdisciplinary and multicentred studies. The second - long term - goal is to make results of this work ( image processing algorithms) easily available to physicians.
Therefore the system has to process a variety of manufacturer- dependent image- formats, videos, non- image data (e.g. ECG's, EEG's..) and conventional patient records.
Data is entered into the system either automatically from imaging entities using the DICOM -communications protocol or interactively using webbrowsers and standard web- protocols like FTP and HTTP.
So scanned photos and grabbed videos can be stored and additional required information (patient- id, study- id ...) can be entered interactively.
All images are converted
into DICOM 3.0 format, which is the only format stored. For studies patient
data will be anonymized.
Queries can be made using a standard webbrowser. Patient- study- or image- specific keywords can be entered. The main advantage of our database system is the integration of image processing routines. The database user can carry out routines like
on the server (using servlets) or java- applets depending on the bandwidth of the network.
In studies many image processing tasks can be performed automatically on a defined set of images. The results like segmented images, 3-D Models, movies but also procedures and parameters can be stored in a DICOM file using private tags.
Hardware and System:
In the first (evaluation)
phase the system is based on a standard SQL- database and webserver on a workstation
with a CD- jukebox.
First tests show that the
system runs stable and facilitates interdisciplinary research. Result in clinical
use will be available in autumn 1998.
At present the system was developed for image- processing researchers, which means that a certain understanding of image processing algorithms and problems is necessary to use it. In future much work will be done to simplify the user-interface.
Corresponding Author:
Dipl.- Ing. Dr. Manfred Gengler
Department of Medical Computer Sciences,
University of Vienna, General Hospital,
Währinger Gürtel 18-20,
A-1090 Vienna/Austria
FAX: +43 1 40400 6667
email: Manfred.Gengler(at)univie.ac.at
Oral presentation at EuroPACS'98, Barcelona, Spain