As David G describes, in theory, PatientID is an LO and Single value Matching is defined to be case sensitive for a Q/R SCP: "This matching is case-sensitive, i.e., sensitive to the exact encoding of the key Attribute Value in character sets where a letter may have multiple encodings (e.g., based on its case, its position in a word, or whether it is accented)" [1].

However, you asked about "programs that touch DICOM files", which is more open ended, and the short answer is that there is no guarantee that anything that messes with an attribute value ("coerces" it perhaps) will recognize/preserve case, even though it might otherwise seem that it should - one could imagine a well-intentioned but misguided import tool "canonicalizing" inbound identifiers to all upper-case, for example.

So I would avoid depending on case in identifiers if I could.

On the subject of hashes in the first place, I worry about them, since may eventually be "broken", not that necessarily is a risk in itself, but with a sufficiently small source space, a brute force attack is a concern. See for example the hash discussion in the MIDI TG report [2][3].

David

1. http://dicom.nema.org/medical/dicom/current/output/chtml/part04/sect_C.2..2.2.html#para_cbd0128a-7ff8-4271-b4e7-e9c379d0a883
2. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC10081345#P269
3. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC10081345#P302

On Tuesday, October 3, 2023 at 5:07:02 PM UTC-4, David Gobbi wrote:
> All DICOM software will preserve capitalization. As for honouring capitalization during a query, case-insensitivity is really only permitted for names (PN). As far as I understand, IDs are always treated as case-sensitive.
>
> https://dicom.nema.org/dicom/2013/output/chtml/part04/sect_C.2.html
>
> But then again, base 36 should be more than sufficient. RSNA-CTP, for example, simply uses base 10 to store hashes. I prefer uppercase and digits in IDs for the rare occasions when I have to recite them over the phone.

On Tuesday, October 3, 2023 at 2:30:17 PM UTC-4, Normand Robert wrote:
> Would nearly all programs that touch DICOM files preserve and honour capitalization if one used an ID hashed to Base62 with the character set a,b,..,z,A,B,..Z,0,1,..,9?
>
> Otherwise I would do base 36 (A,B,..Z,0,1,..,9)
>
> Regards
>
> https://dicom.nema.org/medical/dicom/current/output/chtml/part05/sect_6.2..html

Le jeudi 5 octobre 2023 à 18 h 11 min 25 s UTC-4, David Clunie a écrit :
> As David G describes, in theory, PatientID is an LO and Single value Matching is defined to be case sensitive for a Q/R SCP: "This matching is case-sensitive, i.e., sensitive to the exact encoding of the key Attribute Value in character sets where a letter may have multiple encodings (e.g., based on its case, its position in a word, or whether it is accented)" [1].
>
> However, you asked about "programs that touch DICOM files", which is more open ended, and the short answer is that there is no guarantee that anything that messes with an attribute value ("coerces" it perhaps) will recognize/preserve case, even though it might otherwise seem that it should - one could imagine a well-intentioned but misguided import tool "canonicalizing" inbound identifiers to all upper-case, for example.
>
> So I would avoid depending on case in identifiers if I could.
>
> On the subject of hashes in the first place, I worry about them, since may eventually be "broken", not that necessarily is a risk in itself, but with a sufficiently small source space, a brute force attack is a concern. See for example the hash discussion in the MIDI TG report [2][3].
>
> David
>
> 1. http://dicom.nema.org/medical/dicom/current/output/chtml/part04/sect_C..2.2.2.html#para_cbd0128a-7ff8-4271-b4e7-e9c379d0a883
> 2. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC10081345#P269
> 3. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC10081345#P302
> On Tuesday, October 3, 2023 at 5:07:02 PM UTC-4, David Gobbi wrote:
> > All DICOM software will preserve capitalization. As for honouring capitalization during a query, case-insensitivity is really only permitted for names (PN). As far as I understand, IDs are always treated as case-sensitive..
> >
> > https://dicom.nema.org/dicom/2013/output/chtml/part04/sect_C.2.html
> >
> > But then again, base 36 should be more than sufficient. RSNA-CTP, for example, simply uses base 10 to store hashes. I prefer uppercase and digits in IDs for the rare occasions when I have to recite them over the phone.
> On Tuesday, October 3, 2023 at 2:30:17 PM UTC-4, Normand Robert wrote:
> > Would nearly all programs that touch DICOM files preserve and honour capitalization if one used an ID hashed to Base62 with the character set a,b,...,z,A,B,..Z,0,1,..,9?
> >
> > Otherwise I would do base 36 (A,B,..Z,0,1,..,9)
> >
> > Regards
> >
> > https://dicom.nema.org/medical/dicom/current/output/chtml/part05/sect_6..2.html

Thank you both. At this time I generate a base36 ID with separators like this
SHSC-1Y6AD-YQJOF-ZBVCP-NSKOBNZSAN-WFXFU-90OUY-CN0ZE
so it is easier to read, match and convey to others. We are creating a new system so there is an opportunity to make changes. Adding varying capitalization (using base 62) will also make it harder to visually match to a list. You are confirming that it is not a well used scenario which could therefore reveal occasional problems. Gonna stick with Base32 with separators. I use the same hash used for the name but with a "^" in the middle. Only change might be to shorten both due to complaints that Patient Name and ID overlays in images are long and can mask anatomy unless disabled.
Regards