Reading dicom file

[Hyunsoo-dev]

When i used the code below,

let DicomDict = dcmjs.data.DicomMessage.readFile(dicomFile);

i �got the error as below.

Using multiple character sets is not supported: ,ISO 2022 IR 149

But the standard document says you can put multiple values in a specific character set (0008, 0005).
I'm curious why you limited the number of character set, can you change this restriction?

/// dcmjs.js -> line 10072
if (readInfo.values.length > 1) {
    if (ignoreErrors) {
	    log.warn("Using multiple character sets is not supported, proceeding with just the first character set", readInfo.values);
    } else {
	    throw Error("Using multiple character sets is not supported: ".concat(readInfo.values));
   }
}
readInfo.values = ["ISO_IR 192"]; // change SpecificCharacterSet to UTF-8

[pieper]

There were probably no example datasets available when the code was written or it wasn't a priority for any of the developers/users. It would be great if you could propose a change to make dcmjs more compliant with the standards and provide tests so that further compliance can be tested. It would be good to compare how other toolkits work with this kind of data.

[WillianVarela]

setting the option { ignoreErrors: true } worked for me

[sedghi]

I'm getting this issue also on this dataset that was shared with us in OHIF

1533571122.zip

[salimkanoun]

Find another case similar to this issue, Orthanc accept it but not DCMJS

image_0.zip

[pieper]

Here's a link to the code in question, which explains why ignoring errors will partially work (probably some characters will be garbled).

dcmjs/src/DicomMessage.js

Lines 133 to 160 in 2ca6bb1

	if (cleanTagString === "00080005") {
	if (readInfo.values.length > 0) {
	let coding = readInfo.values[0];
	coding = coding.replace(/[_ ]/g, "-").toLowerCase();
	if (coding in encodingMapping) {
	coding = encodingMapping[coding];
	bufferStream.setDecoder(new TextDecoder(coding));
	} else if (ignoreErrors) {
	log.warn(
	`Unsupported character set: ${coding}, using default character set`
	);
	} else {
	throw Error(`Unsupported character set: ${coding}`);
	}
	}
	if (readInfo.values.length > 1) {
	if (ignoreErrors) {
	log.warn(
	"Using multiple character sets is not supported, proceeding with just the first character set",
	readInfo.values
	);
	} else {
	throw Error(
	`Using multiple character sets is not supported: ${readInfo.values}`
	);
	}
	}
	readInfo.values = ["ISO_IR 192"]; // change SpecificCharacterSet to UTF-8

This comment describes what I understand needs to be done to better handle these datasets: #454 (comment)

[luissantosHCIT]

This can get a bit complex. I think only one character set per level is reasonable. Subsequent levels override the current character set for that level only. Meaning, once your come back to a root level, you have to reset the character set.

Without looking at dcmjs logic deeper than I have, I have a feeling this library would need to keep a stack at hand and track node boundary changes in the header. The DICOM header is structured like a tree with SR nodes being a tree of semi anonymous nodes? and each node can technically switch the character set although I have only dealt with one character switch in practice and it was from latin1 (assumed default) to latin1. Don't ask me why. Haha.

Unfortunately, I do not have the time budget right now to tackle this, but I will be looking forward to finding some time to help with it. Hopefully, someone has the time before I do.

[pieper]

I would think most if not all SR instances would use UTF-8, but I guess technically they could use other character sets.

[luissantosHCIT]

You would think! Haha.

The particular PACS I interact with generates everything in ISO IR 100/ ISO 2022 IR 100/ latin1. I don't know if it has anything to do with the fact it is a Windows only PACS. In retrospect, I would have expected them to generate text in UTF-16, specially because I think some of their code is in the C#/.Net family judging by error messages I encountered in the past.

That is all to say, I do not know why that particular vendor does things the way it does. Often times they do not make sense to me.