[dicom archive] Misleading database representation

[maximemulder]

While using the DICOM archive, I noticed that the field CenterName, in the table tarchive is not nullable. However, this field is often removed when deidentifying patients (such as with DICAT), and as such, it should be nullable in LORIS. The LORIS-MRI scripts currently use an empty string to represent a missing institution, but this is misleading, as it gives the impression that there is always a valid institution while it is in fact optionnal.

I have not checked yet for other fields with the same problem, but it would not surprise me if there are others.

[maximemulder]

This was discussed in today's (2024-04-23) meeting, everyone agrees it would make more sense to make this field nullable instead of using an empty string.