Cannot archive MRI data from Philips scanner, non-secondary or non-parsable files

[Alex Bartnik]

Hello,

Our center recently purchased a Philips MRI scanner and are beginning to use it for new studies, with XNAT supporting the storage and sharing of MRI data collected from this scanner.

For the most part, sending from the scanner to XNAT works, though we have been noticing fairly frequently that XNAT does not correctly parse DICOMs sent from this scanner. We see a warning "WARN-20: Scan XYZ has n non-secondary (or non-parsable secondary) files" for several scans in each session (screenshots attached, our scanning team also does not know where the 0-OT* scans are coming from once XNAT parses the session). Additionally, I can see all of the DICOM files are present on the filesystem in the prearchive (~40,000 files), while XNAT only reports 1 or 2 files for each scan. 

This persists after rebuilding the session in the prearchive or moving it between projects, and also after the session is archived if the warnings are ignored.

I am also attaching the logs for dcmtoxnat within the session's directory in the prearchive, as well as the prearchive.log from the system.

Any help or guidance would be greatly appreciated, this is an urgent issue for us as studies are beginning to ramp up and we need XNAT to reliably share MRI data throughout our center.

We are running a dockerized XNAT v1.8.2.2, build 4.

Thanks!

[Alex Bartnik]

I should add that I spent some time seeing if anything else had reported something similar and found this post and this post, if the contents of those are useful here.

[Charlie Moore]

Hi there,

Are you able to share a fully anonymized example of a study where you see this issue? I don't think we'd be able to tell anything without looking at the data.

Thanks,

Charlie Moore

[Alex Bartnik]

Hey Charlie,

Thanks for the quick response. Unfortunately, we can't upload a full example per our IRB, but I can upload headers if that helps.

I'm uploading two headers, each from the same scan. The first was sorted into a "DICOM" folder by XNAT, the other was sorted into a "secondary" folder by XNAT. FWIW Philips does export some extra data into secondary files, so that adds up. I'm wondering if XNAT gets tripped up by that, but I'm not sure why it wouldn't be able to parse the rest of the DICOM data.

Thanks!

[Alex Bartnik]

The secondary header is 120k lines, I can't upload it. If you'd like to see it I can try uploading to pastebin or similar, but I'm not sure it's worth looking through the whole thing.

[Charlie Moore]

You could try going through Rick's steps on https://groups.google.com/g/xnat_discussion/c/h3r_9AX1Fow/m/4GyBIKFaFQAJ . Could you also send a screenshot of the scan table once the session is archived? 1.8.2.2 is also quite old at this point, so upgrading to the most recent version of XNAT would be a good idea for a variety of reasons. Unfortunately, I'll reiterate that without example data, there's not much I can really do.

Thanks,

Charlie Moore