We are currently in the process of changing PACS vendors. It is known that the data migration is one of the major issues in this process. Does anyone have experience with this?
In our case we want to start with a test migration to determine if the actual migration can be started. Are there any rules or guidelines available how to perform a test migration? What should be included in the test migration? How do we perform the test?
A lot of questions, hope there are some people with experiences that they can share.
Dear Peter, I did this job the last two years at the University Hospital Münster for Cardiology (Angiology. invasive Cardiology and Echo). You can ring me up under ++49 (0251) 83-44903. BW Jörg
We just finished here at UT Southwestern in Dallas. We went from Siemens to Phillips. The migration was contracted out to a database company company named Accuo. What specific concerns do you have?
Where could problems occur? Do you need special focus on certain data? We are now running a test migration and we already ran into several issues that we did not expect (e.g. patient ID issues), how can we ever be sure to tackle all (or at least most) critical problems?
We already migrated several different PACS such as Carestream, Kodak and Global Imaging to DCM4CHEE. There are several problems which may occur as you mentioned. In order to manage them, we developed some applications to firstly scan the metadata of the old PACS in an independent database, secondly validate the metadata, and thirdly run other applications that use the metadata to migrate the exams to the new PACS. Our company can support you. Please let me know is you are interested.
Start your migration in batches. Monitor the error rate. Ideally, It should be in the low single digits. Try and identify why the studies are erring out. When you identify the cause, correct the issue and then you re-send the offending studies to see if they go through. Make sure that the size of your initial batches do not exceed your bandwidth.
It is most important to get the last 2 years of studies onto the system before you take it live. You can migrate studies older than 2 years while you are up on the new system in the background.
One of the issues that we found is that updated entries made to our database (merges, MRN, and name changes) through ADT/ORM messages over the years were kept in the database and not written to the DICOM header until a CGET query came through to pull the images out. To get the most up to date date then you would have to DICOM query each study off of the old database. That would have taken years (literally). One way around this is to extract the database independent of the images. You then generate a new set of order messages with the correct order details (name, MRN, etc.) from the database and send those to your new PACS. The accession number of the new order must match the accession number on the images in the storage. You then send your images over to file under the new orders that you have created matched by accession number.
- Badges
- Science topic
- Similar topics
- Journal Impact Factor