Misalignment when using last‑DICOM alignment & request for BIDS‑JSON support

[angelaaseo]

Dear Lars,

First off, thanks for maintaining PhysIO—our team relies on it a lot!

Setup
• Scanner & sequence: Siemens 3 T (VB17), multi‑echo EPI (3 echoes/TR), multiband = 2
• ECG, Pulse-oximetry, Respiration data: Manual recording via Telnet MPCU (continuous across several fMRI blocks)
• Software: TAPAS Version 6.0.1, PhysIO 2022a, v8.1.0, MATLAB R2020b, SPM12 r6685, Windows 10 (Virtual Desktop)

Issue

When we let PhysIO derive the timing from the last DICOM of each EPI run, the fMRI blocks appear offset relative to the Telnet physiology trace (see attached figure). Switching to the first DICOM eliminates the discrepancy.

Because we terminate each EPI run manually, I suspect the final DICOM’s timing fields (e.g., AcquisitionTime) may not correspond to the actual last TR.
• Is that a known consequence of manual scan termination, or might we be overlooking something else?
• Any other DICOM tags we should double‑check?

Feature request / work‑around

Our data are already BIDS‑converted. The *_bold.json files contain the relevant timing information (RepetitionTime, AcquisitionTime, etc.). Being able to point PhysIO to these JSON files—rather than the first/last DICOM (.IMA)—would simplify workflows.

Is there a known work-around to read timing metadata from BIDS JSON instead of individual DICOMs?

Thanks in advance for any insight!

Best regards,
Angela

[mrikasper]

Dear Angela,

Thank you for providing such a detailed description of the issue and your setup. And happy to hear PhysIO is useful in your research!

Indeed, I believe that stopping the EPI scans manually could lead to the DICOM having the wrong time stamp - which DICOM are you using (do you have one per volume or one per slice?). It should be the last completed volume (and its first slice, if there are multiple DICOMS per volume).

If you use BIDS-converted data (e.g., via dcm2niix), I believe it uses the AcquisitionTime from the first volume, and you should specify this in PhysIO. In our latest release (TAPAS v6.1, PhysIO v8.2), you can put the .json sidecar file for the run as physio.log_files.scan_timing exactly as you suggested.

Please let me know if this works.

Please note that PhysIO is now hosted in its own GitHub repository as part of the new ComputationalPsychiatry GitHub organization. We are still the same TAPAS team, but it should make toolbox-specific development, releases and issue tracking much simpler for everyone.

As part of this transition, please post all new issues here.

Thank you!
Lars