Error in Mediso Dynamic PET Conversion

[QinQikai]

Hi,

I met a problem when I use dcm2niix (v1.0.20240202 and v1.0.20240812) to convert Mediso Dynamic PET DICOM to NIfTI. The output images are in the wrong order. It seems dcm2niix cannot recognize “(0020,9057) InStackPositionNumber” and “(0020,9128) TemporalPositionIndex".

In addition, each DICOM frame has its own “(0028,1053) RescaleSlope”. When I use dcm2niix (v1.0.20220720), it seems only use the first slope value. I am wandering if it was fixed in the latest version.

Thank you!

[neurolabusc]

dcm2niix should be able to use enhanced DICOM ordering. It does handle variable rescale slopes, but this feature is specific to Philips as I have not seen it in use by other manufacturers. If you want me to investigate changing the behavior of the software, you will need to share a concrete example with my institutional email. Alternatively, create a pull request that handles these images.

[neurolabusc]

@QinQikai thank you for the sample images. I have added experimental support for the development branch to dcm2niix. Please validate that this works as expected.

git clone --branch development https://github.com/rordenlab/dcm2niix.gi...
cd dcm2niix/src
make
./dcm2niix /path/to/dicoms
...
Saving as 32-bit float (slope, intercept or bits allocated varies).

I do think you will need to work with the manufacturer to accurately populate the DICOM public tags to allow better BIDS format JSON sidecars. For example "Bq/ml" belongs in the Unit (0054,1001) tag not the Code Value (0008,0100) tag. dcm2niix assumes that DICOM images are truthful and follow the specification.

I do not have access to this hardware, and the provided sample datasets were overwhelmingly large, so please validate that the software works as expected.