Missing echo-1 bold image from heudiconv output on multi-echo scan

Hello Neurostars community,

I’ve recently been running into an issue where, upon running heudiconv on a recent subject with multi-echo scans, the bold.nii.gz image for the first echo winds up missing from the output.

Each subject has 6 fMRI runs, with different permutations of multiband acceleration and acquired echos. In the run with multiband 6 and multi-echo 4 for several of our recent subjects, the first of the four bold images (one for each echo) is missing. Typically, for this echo, the expected output would include mb6me4_echo-1_bold.json and mb6me4_echo-1_bold.nii.gz files, as it outputs for each of the other echos in this run (i.e. 2, 3, and 4). In this case, the initial _bold image (both for .json and .nii.gz) is missing.

The dicoms appear to have the same files as other subjects which have successfully generated this bold image. All contain identical structure and number of files for the multiband 6 multi-echo 4 run. Is anyone familiar with this issue? Any suggestions for solutions or next steps are appreciated. I can provide the heuristics.py code, dicom info, and an example subject’s scans.tsv file (or any other code/info that might elucidate the issue) upon request!

such best troubleshooted when filed as issues against heudiconv if there is suspicion of a bug. Accompany with the log and heudiconv (and dcm2niix) version. In that log would be invocation of dcm2niix. Try calling dcm2niix directly and see what you get (how many files) – may be issue resides within dcm2niix if not within heudiconv.

Thank you Yarik! I’ll post this to heudiconv’s GitHub issues page. For the log, would that just be whatever is printed to the screen when heudiconv runs?
I ran dcm2niix directly and it generated all expected files without issue on a subject that had the missing bold image.

Then smells like a bug in heudiconv indeed, might even be a known one :wink: needs searching but yes – please file an issue with information and link here. Cheers!