Hi all,

I am having a hard time understanding the Part15/Table E.1-1. Application Level Confidentiality Profile Attributes.

I was under the impression there would be an equivalence in between:

De-identification Action Codes D <=> Type 1
De-identification Action Codes Z <=> Type 2
De-identification Action Codes X <=> Type 3

This does not appear to be true for the following case:

Consider "Magnetic Resonance Image IOD", which points to:
* https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_A.4.3.html
Which points to:
* https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.html#sect_C.7.1.1
Which points to:
* https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.html#table_C.7.1.4-1

In other word, the first PatientID attribute is Type 2, while the second one is Type 1 in the following:

[...]
(0010,0020) LO [FOO] # 4, 1 PatientID
(0010,0026) SQ (Sequence with explicit length #=2) # 40, 1 SourcePatientGroupIdentificationSequence
(fffe,e000) na (Item with explicit length #=1) # 12, 1 Item
(0010,0020) LO [BAR] # 4, 1 PatientID
[...]

However now looking at part15:
* https://dicom.nema.org/medical/dicom/current/output/chtml/part15/chapter_E.html#table_E.1-1

All I can see is a 'Z' only:

[...]
Patient ID (0010,0020) N Y Z
[...]

Did I misinterpret the Part15 section, or is this just a typo and the line should read De-identification Action Codes 'Z/D' in this case ?

Thanks

When we introduced the Patient Group Macro to handle multiple small animals being imaged at the same time, I don't think it occurred to us that de-identification was relevant, so the PS3.15 profile was not considered.

Strictly speaking, it would be appropriate to amend the action code as you describe, since Patient ID is Type 1 within the sequences of the Patient Group Macro.

I will submit a CP.

David

On Monday, July 10, 2023 at 10:00:38 AM UTC-4, Mathieu Malaterre wrote:
> Hi all,
>
> I am having a hard time understanding the Part15/Table E.1-1. Application Level Confidentiality Profile Attributes.
>
> I was under the impression there would be an equivalence in between:
>
> De-identification Action Codes D <=> Type 1
> De-identification Action Codes Z <=> Type 2
> De-identification Action Codes X <=> Type 3
>
> This does not appear to be true for the following case:
>
> Consider "Magnetic Resonance Image IOD", which points to:
> * https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_A..4.3.html
> Which points to:
> * https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C..7.html#sect_C.7.1.1
> Which points to:
> * https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C..7.html#table_C.7.1.4-1
>
> In other word, the first PatientID attribute is Type 2, while the second one is Type 1 in the following:
>
> [...]
> (0010,0020) LO [FOO] # 4, 1 PatientID
> (0010,0026) SQ (Sequence with explicit length #=2) # 40, 1 SourcePatientGroupIdentificationSequence
> (fffe,e000) na (Item with explicit length #=1) # 12, 1 Item
> (0010,0020) LO [BAR] # 4, 1 PatientID
> [...]
>
> However now looking at part15:
> * https://dicom.nema.org/medical/dicom/current/output/chtml/part15/chapter_E.html#table_E.1-1
>
> All I can see is a 'Z' only:
>
> [...]
> Patient ID (0010,0020) N Y Z
> [...]
>
> Did I misinterpret the Part15 section, or is this just a typo and the line should read De-identification Action Codes 'Z/D' in this case ?
>
> Thanks

On Tuesday, July 11, 2023 at 1:30:25 PM UTC+2, David Clunie wrote:
> When we introduced the Patient Group Macro to handle multiple small animals being imaged at the same time, I don't think it occurred to us that de-identification was relevant, so the PS3.15 profile was not considered.
>
> Strictly speaking, it would be appropriate to amend the action code as you describe, since Patient ID is Type 1 within the sequences of the Patient Group Macro.
>
> I will submit a CP.

Honestly this was not the answer I was expected :(

Instead something more like this:

---

Hi Mathieu !

Thanks for bringing this back to our attention. You are indeed right Part15 / Attribute Confidentiality Profiles is a bit over-engineered. When Supp 142 was introduced, we thought having multiple action codes for fine tuning ('D' vs 'Z' vs 'X') would make sense for implementers and would help wide spread adoption.

It turns out to be a massive burden to maintain this table as existing attributes are added with different Type as initially planned in Supp 142.

Since we lost the original script to automatically update this table, we are planning to replace all action codes D/Z/X/U combinations with simply two: "X/Z/D" and "X/Z/U*". This should perfectly handle any update of the standard and can be "burned in" any implementation and be future proof while backward compatible.

As a side note you should have already noticed multiple typos like the one you reported, for instance:

* "0008,009C" is not 'Z' but 'X',
* "0008,0050" is not 'Z' but 'X/Z',
* "0400,0550" is not 'X' but 'XD',
* "0400,0552" is not 'X' but 'XD',
* "0040,1001" is not 'X' but 'XD',
* "0010,2299" is not 'X' but 'Z',
* "0010,2297" is not 'X' but 'Z',
* "0040,0009" is not 'X' but XD,
* "0018,1009" is not 'X' but XD,
* so many more but the margin is too small to contain them

Since it does not make much sense to constantly update this table as a favor to implementers, we are planning to globally simplify this table.

Regards
--
Anonymous