Some Dicom format issues

Hi everyone,

I’ve been working with new @itk-wasm/image-io and @itk-wasm/dicom for a while and met some issues with the file format, some of them may be not in plan to be supported, so I would like to know your opinion of them:

  1. The ones which do not have the ‘DICM‘ header info, which are not treat as standard P10 dicom files, if I try to load, there will be some log error to prevent continue, but in 3D slicer they will be displayed, and if i load with dicom-parser, i will get the tag info and the data, I’m not sure if it is not supported because of possible potential legal factors or something else
  2. The series issue, the dicom files from a patient might contain multiple series with axial, coronal and sagittal, if i use readImageDicomFileSeries to read those files, the result could be one of those series(axial, coronal or sagittal) or just console an abort() error even if i changed the param singlesortedseries, do we just plan to support the singlesorted ones?
  3. if i use readImage to read single image Infos and collect them to ge a combined data, the direction i got from image info seems not correct if it is a coronal or a sagittal series(I put the output data to vtk Image and volume to display directly without any other change), I’m not sure if I miss anything important or if I should fix the orientation issue in vtk

All the files above can be parsed with dicom-parser, cornerstonejs will report the ‘DICM‘ header issue but others will be work out, I can add a codeSandyBox to show these issues later if you have interest in them.

@matt.mccormick and/or @jadh4v could comment on itk-wasm plans.

About your 3rd question: VTK did not support direction for a long while, and only added it in recently (version 9 maybe)? I am not sure about vtk.js and vtk-wasm. So you should take a closer look at image direction (orientation) support. If it is not supported, you will need to apply a transform to it, which is taken from itk image’s direction matrix.