Anonymizing DICOM files is a big topic. If you wish to post
information (for example, for programming advice on this group) it is
much safer to start with a non-patient file (e.g. a scan of a test
phantom) which illustrates the same problems rather than trying to
anonymize a real patient file. Even then it may contain information
such as the operator name or facility address.
I once read some detailed advice on anonymizing DICOM files but
unfortunately I can't find that web page any more. There is much more
to be done than just changing the patient name -- for example, the
screenshot at
http://www.santesoft.com/howto/anonymize.html shows the
items that their software can remove or change. Many kinds of DICOM
attributes (including private tags) could contain private information.
Also images can have information "burned into" their pixels.
pydicom has an example program (
http://code.google.com/p/pydicom/
source/browse/trunk/source/dicom/examples/anonymize.py) which changes
or removes some items, but it (like all of pydicom) is released under
the GPL license, so there is no warranty that this will work properly;
you must verify the results yourself. In fact, it has obvious faults
-- it only handles a few kinds of items which might contain sensitive
information, and does nothing about image pixels.
If you google for it, you will find many anonymizer programs. Whatever
method you choose, I suggest inspecting the file directly (including
images) after conversion to make sure there is nothing private
remaining.
And now I must also disclaim this advice, because it is almost
certainly incomplete. Ultimately you must take full responsibility for
verifying there is no private information in a file before publishing
it in any way.
-Darcy