writing dicom to file results in corrupted pixel data

[MiguelSotelo2]

trafficstars

The end goal is to read in the dicom, modify PHI from the header using strings of equal length as those in the tags, and modify the pixel data to remove PHI from pixels I followed the instructions here https://github.com/dcmjs-org/dcmjs/issues/64

The input test image (contains PHI, so I will not upload it) is 97 frames, 600 rows, 800 columns, and 3 samples per pixel

From pydicom, header information is:

(0028, 0002) Samples per Pixel                   US: 3
(0028, 0004) Photometric Interpretation          CS: 'YBR_FULL_422'
(0028, 0006) Planar Configuration                US: 0
(0028, 0008) Number of Frames                    IS: "97"
(0028, 0009) Frame Increment Pointer             AT: (0018, 1063)
(0028, 0010) Rows                                US: 600
(0028, 0011) Columns                             US: 800
(0028, 0014) Ultrasound Color Data Present       US: 0
(0028, 0100) Bits Allocated                      US: 8
(0028, 0101) Bits Stored                         US: 8
(0028, 0102) High Bit                            US: 7
(0028, 0103) Pixel Representation                US: 0
(0028, 0301) Burned In Annotation                CS: 'YES'
(0028, 2110) Lossy Image Compression             CS: '01'
(0028, 2112) Lossy Image Compression Ratio       DS: "0.0"
(7fe0, 0010) Pixel Data                          OB: Array of 6465010 elements
(0002, 0000) File Meta Information Group Length  UL: 208
(0002, 0001) File Meta Information Version       OB: b'\x00\x01'
(0002, 0002) Media Storage SOP Class UID         UI: Ultrasound Multi-frame Image Storage
(0002, 0003) Media Storage SOP Instance UID      UI: 1.2.840.113663.1500.1.313573945.3.2.20190710.111922.593
(0002, 0010) Transfer Syntax UID                 UI: JPEG Baseline (Process 1)
(0002, 0012) Implementation Class UID            UI: 1.2.840.113857.4
(0002, 0013) Implementation Version Name         SH: 'IC2_MC3_470'
(0002, 0016) Source Application Entity Title     AE: 'SHOWCASE_SCP'

In order to rule out anything wrong in my logic, I wrote the following to simply read and write

const arrayBuffer = fs.readFileSync(filename).buffer; 
const DicomDict = dcmjs.data.DicomMessage.readFile(arrayBuffer); 
const dataSet = dcmjs.data.DicomMetaDictionary.naturalizeDataset(DicomDict.dict);
DicomDict.dict = dcmjs.data.DicomMetaDictionary.denaturalizeDataset(dataSet);
let new_file_WriterBuffer = DicomDict.dict.write();
const outputFilename = './test.dcm';
fs.writeFileSync(outputFilename, new Buffer(new_file_WriterBuffer));

I noticed the following:

1. Viewing the input image in Horos, or even in python results in a corrupt pixel data. In Horos the image is displayed well for the first couple of frames. Afterwards, the image is blank.
2. In python, the new header information shows the Pixel Data Array of elements has changed

(7fe0, 0010) Pixel Data                          OB: Array of 6467338 elements

Any help pointing me in the right direction is much appreciated!!

Miguel

[MiguelSotelo2]

In my limited knowledge, I notice that DicomDict.js assumes a specific transer syntax id (explicit little endian) while my file is JPEG Baseline (Process 1) (1.2.840.10008.1.2.4.50)

Thank you, once again!

[pieper]

Hi @MiguelSotelo2 - I haven't used dcmjs with any jpeg image data myself, but I can easily suppose that there are problems. If you can find some public data that can be used to replicate the issue it would be good to add it to the list for future regression testing (for now we are putting datasets here: https://github.com/dcmjs-org/data).

In the meantime, are there other clues to what has changed in the header? Just reading/writing the dataset shouldn't have changed the pixel data that I can think of.

[pieper]

Any follow-up info? Without replication we'll eventually close the issue.