search

Open-Minds-Lab / MRdataset

MRdataset : unified interface to various neuroimaging dataset formats
https://open-minds-lab.github.io/MRdataset/
Apache License 2.0
1 stars 4 forks source link

Validate differences in EchoTimes by using EchoNumbers #2

Open sinhaharsh opened 2 years ago

sinhaharsh commented 2 years ago

Description

The package supports multi-echo modality, but fails to catch discrepancy in echo times. Can we validate in differences in echo times, by using echo number ?

sinhaharsh commented 1 year ago

Gathering evidence, if this is correct way -

Points to consider

  1. Echo Number is optional attribute.
  2. But, it can tell us, if it is an issue or intended

Quotes from dcm2niix github issues

  1. Link

    You may also want to encourage your contacts at Bruker to implement the optional Echo Number (0018,0086) to disambiguate echo times. This allows tools to detect discrete echo times without being concerned about rounding errors with the string based Echo Time (0018,0081) tag.

  2. Link

    As you note, these DICOMs omit echo number (0018,0086) ... [dcm2niix] explicitly inserts the echo time into the filename for multi echo sequences where 0018,0086 is omitted.

  3. Link

    Usually, the output of dcm2niix will provide hints regarding why images from the same DICOM series were saved as separate NIfTI images. For example, variations in slice angulation, echo time, coil, study time, and X-ray exposure are all good reasons to segment data (as these properties can not be encoded in NIfTI).

raamana commented 1 year ago

good catches with these useful pointers/links, Harsh!

let's try parse slice angulation, echo time, coil, study time, and X-ray exposure from DICOM (when available) and keep them in MRdataset for now, so we can decide how to handle them in future.

again, let's table this for discussion in redesigning MRdataset 2.0 as it may involve more testing as we change some key things