read dicom with jpeg compress data

[chenkaiC4]

hi, i want to read a dicom with data compressed as jpeg. i find the daikon.js, and use it to get uncompress dicom image, but then, how to use itk-js to read the image, because i use vtk.js too, i know vtk.js depend on itk.js to read such as image, dicom etc.

main code logic is below:

import itkreadImageDICOMFileSeries from 'itk/readImageDICOMFileSeries';
import vtkHttpDataAccessHelper from 'vtk.js/Sources/IO/Core/DataAccessHelper/HttpDataAccessHelper';

vtkHttpDataAccessHelper.fetchBinary(url).then((arrayBuffer) => {
    const image = daikon.Series.parseImage(new DataView(arrayBuffer));
    if (image === null) {
        console.error(daikon.Series.parserError);
    }

    const rawData = image.getRawData(); // ArrayBuffer

    const filename = `test.dcm`;

    const file = new File([rawData], filename);

    itkreadImageDICOMFileSeries(null, [file])
        .then(({ image: itkImage, webWorker }) => {
            // code not exec here
            // get a Error: Could not read file: /work/test.dcm
        })
})

update:

because the raw image data have got, i think i have to create a Image instance, but need a correct ImageType.

https://github.com/InsightSoftwareConsortium/itk-js/blob/f2b16ac4b9097367969dc1ff961037bc43441753/src/Image.js#L4-L23

https://github.com/InsightSoftwareConsortium/itk-js/blob/f2b16ac4b9097367969dc1ff961037bc43441753/src/ImageType.js#L4-L12

but i don't know how to determine the componentType componets and other input args of constructor ImageType.

thanks for your suggestions!

[thewtex]

@chenkaiC4 you do not need to use daikon -- use itk.js directly and pass it the array buffer.

Here is an example:

https://github.com/Kitware/itk-vtk-viewer/blob/ddc38926657cadfa38280ab4da4e45ca7c482a61/src/processFiles.js#L15-L24

[chenkaiC4]

@thewtex Thanks for your suggestion. But my dicom is compressed to JPEG, itkreadImageDICOMFileSeries can't read it. I have try it, with error 'Could not read file: xxx.dcm'.

[thewtex]

@chenkaiC4 I reproduced the read failure with the daikon.js test dataset -- I will see about fixing the ITK build so JPEG2000 compression is supported.

[chenkaiC4]

@thewtex 👍 Native support by itk.js is the best solution. Glad to see it 😃

[chenkaiC4]

@thewtex hi, i have create a Image instance manually which declared at https://github.com/InsightSoftwareConsortium/itk-js/blob/6377e1890d524fa9cf688c21698cfa4cfb6ca02b/src/Image.js#L4 , the raw arraybuffer is get from the daikon.js's concatenateImageData method which will reorder the 2D images of dicom and concat them.

relative code shows as follow, it works. But the slicer mode display is not correct. see the two picture.

      const itkImage = new Image(new ImageType(3, dataType));
      itkImage.size[0] = row;
      itkImage.size[1] = col;
      itkImage.size[2] = dataLength;
      itkImage.spacing[0] = pixelSpacing[1];
      itkImage.spacing[1] = sliceThickness / 2;
      itkImage.spacing[2] = pixelSpacing[0];
      itkImage.direction.setElement(0, 0, 0.0);
      itkImage.direction.setElement(0, 1, -0.0);
      itkImage.direction.setElement(0, 2, -1.0);
      itkImage.direction.setElement(1, 0, 1.0);
      itkImage.direction.setElement(1, 1, 0.0);
      itkImage.direction.setElement(1, 2, 0.0);
      itkImage.direction.setElement(2, 0, 0.0);
      itkImage.direction.setElement(2, 1, -1.0);
      itkImage.direction.setElement(2, 2, 0.0);
      itkImage.data = new Uint16Array(file);

      const imageData = vtkITKHelper.convertItkToVtkImage(itkImage);

Follow picture is well display, it display a dicom series not compressed. but the second picture is wrong, the dicom series is the same with the first picture's. i do this just want to test the combination with daikon.js and itk.js. So i use a not compressed data to test the code. From the two pictures, the spacing and direction seemd to be wrong. but i don't know how to fix it.

[thewtex]

@chenkaiC4 well done. Eventually we will get JPEG2000 support in itk.js, but this can work as a stop-gap.

You do not want to use sliceThickness for the third axis spacing. This refers to the imaging thickness (think Point Spread Function) of the slice as opposed to how it is sampled. To find the value, look at the PixelSpacing or difference between the ImagePositionPatient of the first two slices. The direction is related to the ImageOrientationPatient. There are related references online.

[chenkaiC4]

@thewtex thanks, i will have a try to fix it. 👍

[awwabi]

Hi @chenkaiC4 can I email you to ask a lot things about dicom reader?
I need your help

[chenkaiC4]

@LabbaAwwabi chenkaiC4@email.com i'm not sure i can solve your problems, just have a try.

[martinweihrauch]

Hi!
I just found this thread via google!

What I am trying to accomplish: I have JPEG-LS compressed DICOM images and want to show them in a browser with a certain windowing, which I apply to it.

Do you have - by any chance - any simple code which does that in JS. I. e., start with a DICOM image (e. g. by link) and result in the canvas image with the correct windowing applied? Also stating, exactly which JS libraries to include? I do not need JPEG2000, only JPEG-LS.

Best regards and thanks a lot!

Martin

[engrussman]

@martinweihrauch Did you get any solution for JPEG-LS?

[juanprietob]

Workaround using gdcm-js
This is not ideal as it takes double the amount of time to process a single image.

const fs = require('fs');
const gdcm = require("gdcm-js");
const readImageLocalFileSync = require('itk/readImageLocalFileSync.js');

var dicomoriginal = new Uint8Array(fs.readFileSync('compressed.dcm'));
var memfs = [{name: "input.dcm", data: dicomoriginal}]; 
var args = ["-i", "input.dcm", "-o", "output.dcm", "-w"]; 

var result =  gdcm.gdcmconv({MEMFS:memfs,arguments:args});

fs.writeFileSync('temp.dcm', result.MEMFS[0].data)

var in_img = readImageLocalFileSync('temp.dcm');
console.log(in_img)

[juanprietob]

This reader available in npm med-img-reader reads compressed DICOM images and returns in the ITK format so it is compatible with VTK.js as well.

[thewtex]

Another option here is to add optional support for readImageDICOMSeries to use ITK/GDCM, which does have JPEG2000 support (it is currently using ITK/DCMTK without JPEG2000 support).

[thewtex]

itk/readImageFile and itk/readImageDICOMSeries use GDCM under the hood since 12.1.0 (not. that readImageDICOMSeries's interface changed recently), so these issues should be addressed.