Re: medical image database

From: Dave Kimmel <criscokid_at_v-wave.com.nospam>
Date: 17 Aug 2000 20:06:13 GMT
Message-ID: <slrn8pohdj.loe.criscokid_at_deepcore.gsm>


In article <399C21FE.F3C3C8C2_at_v1.wustl.edu>, James Dickson wrote:
>
>I am currently designing a database with Sybase to store large
>quantities of Magnetic Resonance Image data and numerous experimental
>data files. These files are currently residing on the filesystem and
>generally consist of an attribute based header and raw data . I am
>trying to decide whether to insert this data into the database as
>text/blob data or insert the attribute data from the header and store
>the raw data in the filesystem.

This reminds me of something I did recently. I have written a system to store the policy and procedure manual at work. I initially took the approach of storing the documents out in the filesystem and the "meta data" (date created, parent document, keywords, title, etc) in a database.

The problem I found with doing this was that it didn't lend itself well to making the system better. I had to make the system store old versions of documents, so I either had to do some nasty hacks in my code to make this work or store them in the database and use some less nasty hacks in the code.

After this experience, my vote goes to storing them as BLOBs if the database can handle it so that you save yourself a lot of time and effort. If the database can't handle the amount of data you want to store (my transition from storing documents in the filesystem to storing them in the database required a switch from PostgreSQL to MySQL due to a row size limit in PostgreSQL), either consider either using a different database or storing them in the filesystem.

Hope that helps
-- Dave Kimmel

   criscokid_at_v-wave.com
   ICQ: 5615049 Received on Thu Aug 17 2000 - 22:06:13 CEST

Original text of this message