Skip to content

Latest commit

 

History

History
88 lines (57 loc) · 3.99 KB

README.md

File metadata and controls

88 lines (57 loc) · 3.99 KB
Raw

Streamer for lab dataflow

This package implements a data streamer service allowing an automatic dataflow from the scanner consoles to the project storage and RDM collections.

It is currently implemented for the MRI and MEG labs at the DCCN. The TG is now also working on an implementation of a standard dataflow for the other labs and data types (e.g. behavior, eye tracking, etc.).

Most of the documentation that you find below is for system administrators rather than for researchers.

Introduction

The service consists of two components, the redis database and the streamer server. The two components are dockerised, and orchestrated with the docker-compose.

The streamer server is implemented as a NodeJS web application based on the express framework.

Build and run the service

  • prepare the config directory, and copy the example from docker/streamer/config/default.conf
  • in the directory of docker, run docker-compose build --force-rm
  • run the script docker/redis/docker_host_config.sh to adjust host's kernel parameter, for the performance of the redis database.
  • start up the service by docker-compose run -d

Submit streamer job (an example of MEG)

The following cURL command shows how to trigger a dataflow concerning data in the directory /ctfmeg/odin/data/meg/ACQ_Data/20161121/ on the MEG console.

$ curl -X POST -u admin http://{streamer_hostname}:3001/meg/20161121

Note that the prefix /ctfmeg/odin/data/meg/ACQ_Data/ is skipped as it is already provided in the configuration file as the value MEG.consoleDataDirRoot.

Submit streamer job (an example of MRI)

The streamer job for MRI is provided in the basis of a DICOM series. Normally, it is called by the Orthanc PACS server when a series is considered as "stable" (i.e. the series is considered as stable if the Orthanc server doesn't receive any update on it for one hour).

The following curl command activates the data flow on a DICOM series identified by the Orthanc server as 1f3df579-b58352b8-923b6bd9-e44aefa4-21581e31.

$ curl -X POST -u admin http://{streamer_hostname}:3001/mri/series/1f3df579-b58352b8-923b6bd9-e44aefa4-21581e31

Extending the service for other modality

The modality plugin is located in the docker/streamer/lib directory, and it should be named as modality<MODALITY_TYPE>.js. The <MODALITY_TYPE> should also be reflected in the configuration file docker/streamer/config/default.json (see below).

The modality plugin is required to export two functions and one JSON object.

The JSON object should be exported as restPaths. It defines the REST interface of creating a streamer job for the plugin.

var restPaths = {
    'postJob': '/:date/:ds?'
};
module.exports.restPaths = restPaths;

The two required functions are:

  • createStreamerJob: function to convert RESTful request into streamer job
  • execStreamerJob: function to process the streamer job

An example for a modality of type TEST is given as modalityTEST.js.

Use the modality plugin

To make use of the modality plugin, one should define a modality instance in a streamer's configuration file (docker/streamer/config/default.json) under the Modalities sector. Define an unique name of the instance as a key in the sector, and a corresponding value containing at-least an attribute called type referring (case sensitively) to the plugin. Once the streamer is (re-)started, a new REST interface prefixed with the name of the modality instance will be created for accepting new streamer jobs to be processed by the plugin.

The following example creates a new modality called myTest, using the modalityTEST.js as the plugin:

{
    ...
    "Modalities": {
        "myTest": {
            "type": "TEST"
        }
    }
}

and the following REST interface for creating new job to be processed by the modalityTEST.js plugin:

  • http://<streamer_host>:<streamer_port>/myTest/:date/:ds?

where the part /:date/:ds? is defined by the JSON object restPaths exported by the plugin.