dicomParser skips PixelData if it can not determine the length of an unknown tag about dicomparser HOT 7 CLOSED

I don't think that's the fix we want. I'm actually not sure how dicomParser is even getting into findItemDelimitationItemAndSetElementLength. It shouldn't be using that on the SQ element, only on individual items. Do you know?

It's hard to tell from just the dump, but it looks like that item has an explicitly-defined length of 8 bytes, which wouldn't even cover the (0507,0000) element. If that's the case, per the standard there wouldn't even be an item delimitation item and dicomParser should be throwing an exception when it looks for a second item and doesn't find the item tag (see readSequenceItem). That would mean the file isn't standard-compliant.

from dicomparser.

This sounds like an invalid DICOM file, especially if anonymization tools remove the tag. Your fix may solve your problem, but it would make the library non standards compliant so is not an acceptable change for the library. My general philosophy is to keep the library standards compliant but provide hooks when possible for higher levels to implement workarounds for non standard data. Unfortunately I can't really suggest any hooks for a workaround without looking at the file itself

from dicomparser.

This is the (0507,1001) element it reads:
{
"tag": "x05071001",
"vr": "UN",
"length": 4294967295,
"dataOffset": 2384
}

Its VR is not SQ and it has undefined length, so that is why it is getting in findItemDelimitationItemAndSetElementLength.

from dicomparser.

Rereading the DICOM standard on undefined length handling and it looks like we have the logic wrong for non SQ elements. Use of undefined length appears to be restricted to specific VR's and is also parsed in the same way as a SQ element. I need to find some data sets that have this condition so I can verify a fix, if anyone has an anonymized one please send it my way :)

from dicomparser.

Just pushed version 1.7.3 which should fix this - can you please test and confirm? Your workaround fix is not correct

from dicomparser.

actually test 1.7.4, a bug slipped into 1.7.3 on accident :(

from dicomparser.

1.7.4 works fine for that case

from dicomparser.