1
votes

My question is: suppose I have a medical modality, for example a MRI, after the scanning for a patient, it will generate the reconstructed dicom data, and I think I have two solutions to save the dicom data,

One is to save the dicom data as dicom files and then save the files into a file folder structure system; the other is to save the dicom data into a database system (For example: there may be a patient table

which is to store patient information, and there may be a study table which is to store study information and etc). I think the database solution is better than a file folder structure system (What is your opinion?).

And if I use the database solution, then when choose a database, such as Berkeley DB or MS SQL Server or Oracle or other DB, which database system should I choose? Is there any special consideration for

medical usage? What is the criteria for the DB selection?

Can anyone give me some suggestions?

Thanks.

1
Hi Evan, thanks for your reply. I have a further question, why you select PostgreSQL but not other DB system, such as MySQL, SQLLite and etc? What consideration do you have when you decide to use PostgreSQL? Can you explain that? What I know about the PostgreSQL is that PostgreSQL is totally free no matter what usage.jealtw

1 Answers

3
votes

The debate over storing images on the file system vs in a database is well-worn; reference other StackOverflow questions (examples here and here).

Before you roll your own system, I'd encourage you to look at an imaging platform called XNAT (disclosure: I'm one of the developers). Our platform handles DICOM, among other imaging formats, out-of-the box. It can also act as a C-STORE SCP (meaning you can send data to it directly from a scanner). XNAT is open source and freely available. If your venture is commercial/clinical, there are also forks of the project that operate in those arenas.

For what it's worth, our system stores the image metadata in a database (PostgreSQL) and the DICOM files themselves on the file system.