dicom-sr-parser

Extract TEXT (0040,A160) from DICOM Structured Reports by Concept Name Code Value (0008,0100) — minimal Node API built on dicom-parser.

npm install dicom-sr-parser

Requires Node.js 18+

Quick start

CLI

npx dicom-sr-parser report.dcm QURE_CODE_12
npx dicom-sr-parser --all --json report.dcm TEST_CODE_01
npx dicom-sr-parser --coding-scheme 99VENDOR report.dcm QURE_CODE_12

| Flag | Description | |------|-------------| | --all | All matches (uses parseAll) | | --json | JSON on stdout | | --coding-scheme <id> | Filter (0008,0102) | | --code-meaning <text> | Filter (0008,0104) | | -h, --help | Usage |

Default codeValue: QURE_CODE_12

Library

const fs = require("fs");
const { parse, parseAll, parseFromFile } = require("dicom-sr-parser");

const buffer = fs.readFileSync("report.dcm");

const { found, text } = parse(buffer, "QURE_CODE_12");
const all = parseAll(buffer, "QURE_CODE_12", {
  codingSchemeDesignator: "99VENDOR",
});
const fromDisk = parseFromFile("report.dcm", "TEST_CODE_01");

Long impressions (TEXT / UT): by default strings up to 1MB are read (maxElementLength; dicom-parser alone truncates at 128).

parse(buffer, "IMPRESSION", { maxElementLength: 2 * 1024 * 1024 });

ESM

import { parse, parseAll } from "dicom-sr-parser";

TypeScript

Types ship with the package (src/index.d.ts). Run npm run test:types.

dcmjs (optional)

npm install dcmjs

const { parseNaturalizedDataset, parseWithDcmjs } = require("dicom-sr-parser/dcmjs");

parseNaturalizedDataset(naturalizedDataset, "126000");
parseWithDcmjs(buffer, "QURE_CODE_12");

Browser

npm run build

import { parse } from "dicom-sr-parser/browser";

See examples/browser-demo.html and docs/guide.md.

Vendor presets

templates/qure.js, templates/rsna.js, templates/tid3110.js

API reference

| Export | Description | |--------|-------------| | parse(buffer, codeValue, options?) | First match → { found, text, codeValue } | | parseAll(buffer, codeValue, options?) | All matches → { found, matches: [{ text, path }], codeValue } | | parseFromFile(path, codeValue, options?) | Same as parse | | findTextByCodeValue(items, codeValue, options?) | Walk pre-parsed Content Sequence | | findAllByCodeValue(items, codeValue, options?) | All matches on pre-parsed tree | | findTextByCodeMeaning(items, codeMeaning, options?) | Match (0008,0104) only | | matchesConceptCode(codeItem, codeValue, options?) | Low-level concept matcher | | extractTextFromContentItem(item) | Read TEXT/UT from one content item | | dataSetToContentTree(dataSet, options?) | explicitDataSetToJS → Content Sequence | | TAG | Hex tag keys used internally | | DEFAULT_MAX_ELEMENT_LENGTH | Default maxElementLength (1MB) |

path in matches uses 1-based ordinals along nested (0040,A730) (DICOM Referenced Content Item Identifier style).

dicom-sr-parser vs dcmjs

| | dicom-sr-parser | dcmjs | |--|----------------------|-----------| | Size / deps | Tiny, dicom-parser only | Large, full DICOM toolkit | | Goal | Get labeled SR text by code | Read/write SR, SEG, TID 1500, OHIF | | API | parse(buffer, "CODE") | Naturalized JSON, derivations | | Best for | ETL, scripts, vendor codes | Web viewers, derived objects |

Use dcmjs when you need standards-first SR creation or OHIF integration. Use dicom-sr-parser when you only need one field from an SR in Node.

Examples

node examples/node-etl.cjs ./dicoms
node examples/import-esm.mjs report.dcm TEST_CODE_01

Development

npm install
npm run build
npm test
npm run test:types

Roadmap

ROADMAP.md

Changelog

CHANGELOG.md

License

MIT