mrconvert: enhanced 4D DICOM encoding formats are not fully supported #2690
Description
Firstly, I apologise for the perhaps vague information in this bug report - DICOM encoding is far from my area of expertise!
Some 4D enhanced DICOM encoding formats are not supported by MRtrix3. Specifically, we have a 4D diffusion tensor image (from a Bruker scanner) that can be read and interpreted by MRtrix3 without issue. The functional group sequence encoding uses Item/SequenceDelimitationItems with undefined length. Upon upload to our DICOM server, this undergoes a re-encoding to sequences with explicit length. This is a valid DICOM format and can be viewed/opened by, for example, FSL. However, when opening with mrview, the images are corrupted (just noise), the axis dimensions are converted from [X,Y,Z,B] to [X,Y,1,Z*B], and the B-tables are truncated to just a single row.
ORIGINAL IMAGE:
trosenow@mrfellowpc:~/Downloads/tmp3$ mrinfo /data/research/chemobrain_mouse_43_chemobrain_mouse_43_chemobrain_E6_P1_EnIm1.dcm
mrinfo: [done] scanning DICOM folder "/data/rese..._chemobrain_E6_P1_EnIm1.dcm"
mrinfo: [100%] reading DICOM series "DTI_2seg_31dir b1/2000 0.2/0.4"
************************************************
Image name: "chemobrain_mouse_43^^^^ (chemobrain_mouse_43) [MR] DTI_2seg_31dir b1/2000 0.2/0.4"
************************************************
Dimensions: 100 x 52 x 44 x 67
Voxel size: 0.2 x 0.2 x 0.4 x ?
Data strides: [ -1 -2 3 4 ]
Format: DICOM
Data type: unsigned 16 bit integer (little endian)
Intensity scaling: offset = 0, multiplier = 33.769885284499999
Transform: 0.9968 -0.07876 -0.01592 -9.056
0.07725 0.9938 -0.08014 -7.251
0.02213 0.07865 0.9967 -9.642
FlipAngle: 90
PixelBandwidth: 3000
RepetitionTime: 3
comments: chemobrain_mouse_43^^^^ (chemobrain_mouse_43) [MR] DTI_2seg_31dir b1/2000 0.2/0.4
DOB: 20/06/2022
DOS: 20/06/2022 13:52:17
dw_scheme: 0,0,0,68.374951296379706
[67 entries] 0,0,0,68.374951296379706
...
-0.47919169244646398,-0.87310715743715195,-0.089773122493923593,1027.969034393675
-0.47919169244646398,-0.87310715743715195,-0.089773122493923593,2025.5518261924087
DOWNLOADED IMAGE
trosenow@mrfellowpc:~/Downloads/tmp3$ mrinfo serverver/2.16.756.5.5.200.8323328.68896.1655774923.34-60001-1-1tl757z.dcm
mrinfo: [done] scanning DICOM folder "serverver/...4923.34-60001-1-1tl757z.dcm"
mrinfo: [100%] reading DICOM series "DTI_2seg_31dir b1/2000 0.2/0.4"
************************************************
Image name: "chemobrain_mouse_67^^^^ (chemobrain_mouse_67) [MR] DTI_2seg_31dir b1/2000 0.2/0.4"
************************************************
Dimensions: 100 x 52 x 1 x 2948
Voxel size: 0.2 x 0.2 x 0.4 x ?
Data strides: [ -2 -3 4 1 ]
Format: DICOM
Data type: unsigned 16 bit integer (little endian)
Intensity scaling: offset = 0, multiplier = 41.234815597299999
Transform: 0.9999 0.006102 -0.01069 -9.909
-0.009416 0.9385 -0.3452 -6.514
0.007925 0.3452 0.9385 5.889
FlipAngle: 90
PixelBandwidth: 3000
RepetitionTime: 3
comments: chemobrain_mouse_67^^^^ (chemobrain_mouse_67) [MR] DTI_2seg_31dir b1/2000 0.2/0.4
study: chemobrain_mouse_67_chemobrain [ ORIGINAL PRIMARY DIFFUSION NONE ]
DOB: 21/06/2022
DOS: 21/06/2022 10:50:25
dw_scheme: -0.47919169244646398,-0.87310715743715195,-0.089773122493923593,2025.5518261924149
To Reproduce
- Obtain a 4D enhanced dicom from a Bruker scanner (or maybe others?).
- Upload to any XNAT server (or other dcm4chee server).
- Download the re-encoded file
- View with mrview or convert with mrconvert.
The original scan and the corrupted server version can be found at this link:
https://filesender.aarnet.edu.au/?s=download&token=353ea827-b674-4493-b876-abbb006663d3
Further discussion on the XNAT forum can be found here:
https://groups.google.com/g/xnat_discussion/c/8HvBDU1GT7c
Platform/Environment/Version
Please provide the following information:
-
OS: Ubuntu 22.04.2 LTS
-
MRtrix3 version: == mrinfo 3.0.4-46-g8a192720 ==
See relevant DICOM standard document here: https://dicom.nema.org/medical/dicom/current/output/html/part05.html#sect_7.5