Dimon mosaic problem

AFNI Message Board
1 
afni --version
Precompiled binary linux_openmp_64: Aug  5 2026 (Version AFNI_26.2.03 'Gordian III')

Dear Colleagues,

Hi! I ran this

Dimon \
-infile_pattern ${CDIRDCM}/"*.dcm" \
-dicom_org \
-GERT_Reco \
-gert_create_dataset \
-gert_write_as_nifti \
-gert_outdir ${CDIROUT} \
-gert_to3d_prefix ${CFILBAS} \
-no_wait

command twice, once in 2024 and once this week, on very similar .dcm files. It generated these

to3d -prefix sub-00005_ses-1_task-sleep_run-20240214_1939.nii -time:zt 40 2700 2.0sec FROM_IMAGE -@
to3d -prefix sub-00007_ses-1_task-sleep_run-20260722_1715.nii -time:zt 1 1200 3.0sec alt+z -@

commands. The former has the correct number of slices, 40, whereas the latter has an incorrect number of slices, 1. This

dicom_hdr /raid/common/projects/epilepsy/sourcedata/sub-00007/ses-1/func/2-amri_ep2d_bold_sleep3/2.113564.9.1.3135784786.90.2.5001194543-2-1-1b0qck1.dcm
...
0002 0002       28 [166     ] //META Media Stored SOP Class UID//1.2.840.10008.5.1.4.1.1.4.1
0028 0008        2 [3006    ] //           IMG Number of Frames//50
...

shows that the .dcm file uses the standard UID for Enhanced MR Image Storage, though I do not understand that, and the correct number of slices/frames: 50. I was able to assemble the .dcm files successfully using dcm2niix. Adding -assume_dicom_mosaic had no effect. What do you think?

Sincerely,

Dante

2

Dear Colleagues,

Hi again. I am not sure if this thread should be closed because I probably missed something, but, to keep you informed, I have switched the lab's code from Dimon to dcm2niix and am moving forward. I hope my switch does not cause any problems later in our pipelines, does not cause worlds to collide, does not cause the fires of Mount Doom to awaken, and does not cause shadowy, cloven-hoofed demons to drag me to Hell upon my death.

Sincerely and In Good Humor,

Dante

3

This sounds good. Dimon does not read "enhanced" dicom yet, and the dcm2niix_afni does. This may not be good for realtime processing, but for regular conversion, it's fine. Dante's inferno is all yours, but you should be good.

4

Hi Dante,

I was going to contact you and verify whether those were enhanced, but it should be mostly fine to move forward with dcm2niix[_afni]. One caveat is that it reorients the data by default, to RPI I believe.

-rick