Bruker DICOMs
DICOM files from the Bruker 9.4T system are of the Enhanced MR information object type. This allows for storing multiple frames (slices) into one file.
Nifti conversion
Nifti conversion tools are designed for “non-enhanced” MR DICOMs, and may not work well with Enhanced MR DICOMs. Enhanced MR DICOMs can be converted with an up-to-date version of dcm2niix. Ensure that your version is at least v1.0.20181125 and download the latest version if necessary.
If the output Nifti files are incorrect, please email Igor Solovey or create an issue directly on the rordenlab/dcm2niix repository.
Workaround
Only use this as a last resort in case an up-to-date version of dcm2niix is not working!
In order to convert Enhanced MR DICOMs to Nifti, you can first convert them to single-frame MR information object type first.
-
Download the
dcm4chetoolbox and extract. -
Either use a bash script provided in $12 or follow these steps:
i. Convert an Enhanced MR image DICOM to a series of single-frame MR image DICOMs:SINGLE_FRAME_OUTPUT_DIR=/path/to/data/single-frame PATH_TO_ENHANCED_MR_DICOM=/path/to/data/enhanced.dcm /path/to/dcm4che/bin/emf2sf --out-dir ${SINGLE_FRAME_OUTPUT_DIR} --not-chseries ${PATH_TO_ENHANCED_MR_DICOM}ii. Convert as before using
dcm2niixormri_convert:NIFTI_OUTPUT_DIR=/path/to/nifti/directory dcm2niix -o ${NIFTI_OUTPUT_DIR} ${SINGLE_FRAME_OUTPUT_DIR}
Known Issues
- Rescale Slope (0028, 1053) and Rescale Intercept (0028, 10552) are known to be wrong in Enhanced MR DICOM files from Bruker. The wrong value is intentional because Bruker wants to users to use Real World Mapping Macro (0040,9096) values instead. Unfortunately Bruker forgot to add this new field to the Enhanced MR DICOM Instances so they just unintentionally broke scaling. You can however extract the required information from DICOM Tag (0177, 1101) which contains the Bruker
visu_parsfile contents. The$VisuCoreDataSlopeand$VisuCoreDataOffsvariables have the required slope and intercept, respectively. - DICOM diffusion values, in particular (0018, 9602) to (0018, 9607), are known to be incorrect. The actual diffusion values can be found in DICOM Tag (0177, 1100) which contains the Bruker
methodfile contents. The$PVM_DwBMatvariable contains the required information. It is important to note that the diffusion in$PVM_DwBMatis in Read, Phase, Slice coordinates. This is different than the DICOM XYZ patient coordinates.
Tools
-
https://git.cfmm.uwo.ca/bruker/pybruker contains Python tools for parsing
visu_parsandmethodfiles. These files are in JCAMP DX version 4.24 format. -
https://git.cfmm.uwo.ca/baron-lab/baronAnalysis contains a Matlab script for computing .bval and .bvec tables from the enhanced DTI DICOMs.