[Neuroimaging] JSON-LD and DICOM?

Ian Malone ibmalone at gmail.com
Fri Jun 30 11:41:40 EDT 2017

On 30 June 2017 at 14:32, Matthew Brett <matthew.brett at gmail.com> wrote:
> Hi,
> On Fri, Jun 30, 2017 at 2:25 PM, Andrey Fedorov
> <andrey.fedorov at gmail.com> wrote:
>> Hi guys,
>> no, I don't think that file will be useful to you.
>> For us, the plan was to utilize that file as a central location to
>> link to the DICOM standard text and attribute definitions, but it has
>> not yet become a priority to organize this. We are focused on the
>> functionality and bug fixes for the library.
>> If you come up with something, I would be curious to see, although I
>> don't think that copying DICOM attributes to a NIfTI header is a good
>> idea.
> That's interesting - why do you not think it's a good idea?

I'd be quite interested to see something like this, it would be useful
for processing pipelines where information that is present in the
DICOM files is required that currently needs to be supplied
out-of-band. The bandwidth per pixel attribute for example is needed
when applying susceptibility correction to EPI images from Siemens MRI
data, or even just the field strength attribute for MRI can be useful.
On the downside, DICOM is quite complex and used differently by
different vendors and modalities, so there can be multiple values of
attributes for series, study and patient levels, and different fields
that can be used - not all vendors use that bandwidth per pixel
attribute for EPI MRI data. Then you've got private blocks (can be
useful information, but very vendor idiosyncratic and another layer of
structure) and anonymisation to worry about. (Converting to NIfTI is a
good way to start cleaning identifiable information, since it leaves
you only a couple of fields that could contain anything identifiable
and almost certainly just contain the conversion software label.)
So... problems, but could be cool.


More information about the Neuroimaging mailing list