Lightweight DICOMweb Server with CouchDB
Note: this is a work in progress and not intended for production or clinical use.
More background information can found in https://na-mic.github.io/ProjectWeek/PW30_2019_GranCanaria/Projects/DICOMweb-CouchDB/
The dicomweb-server is a fastify server that speaks DICOMweb to clients and fullfills their requests using CouchDB or other plugin services.
By default, the authentication is none and the application mode is development. You can change the authentication method by changing the auth attribute in config/development.js The value you put in should be the name of a json file in the config directory. A sample config for authentication should have the following information
{ "realm": "your-realm", "authServerUrl": "your-auth-server-port-and-port", "clientId": "your-client-id", "clientSecret": "your-secret" }
If using the default authentication of couchdb with an admin account, you will need to specify the admin username and password in config/development.js in the below style:
{ dbServer: process.env.DB_SERVER || 'http://username:password@localhost' }
git clone git://github.com/dcmjs-org/dicomweb-server
cd dicomweb-server
npm install
Install CouchDB.
Initially your CouchDB database starts empty, but dicomweb-server will set up the internal database and design documents so there is no need to configure it.
You can run tests by running npm test
.
Be sure to have CouchDB running at localhost:5984 (the default), then start the dicomweb-server:
npm start
The server should be ultimately compatible with any DICOMweb client library.
We test with a Python implementation dicomweb_client.
Get study list:
dicomweb_client --url http://localhost:5985 search studies
Store a DATA_DIRECTORY of DICOM image files (here with the ".IMA" extension). Adjust the command line to match the location and naming of your files. (The -n25
option to xargs is for batching files, leading to fewer calls and thus less overhead.)
find DATA_DIRECTORY -iname \*.IMA -print0 | xargs -0 -n25 dicomweb_client --url http://localhost:5985 store instances
It's possible to use this server as a backend to the OHIF Viewer using a configuration like this. (See this file).
const dicomweb_serverConfig = {
routerBasename: "/ohif",
rootUrl: "http://localhost:2016/ohif",
servers: {
"dicomWeb": [
{
"name": "dicomweb_server",
"wadoUriRoot": "http://localhost:5985",
"qidoRoot": "http://localhost:5985",
"wadoRoot": "http://localhost:5985",
"qidoSupportsIncludeField": true,
"imageRendering": "wadouri",
"thumbnailRendering": "wadors",
"requestOptions": {
"requestFromBrowser": true
}
},
]
}
};
Note that currently the imageRendering
option must be wadouri