NA-MIC Project Weeks

Back to Projects List

Lightweight DICOMweb Server with CouchDB

Key Investigators

• Emel Alkim (Stanford)
• Steve Pieper (Isomics)
• Hans Meine
• Erik Ziegler

Project Description

This project aims to build on prototype lightweight DICOM server that was developed last year

Objective

• Develop easy to install and test recipes / scripts
• Compare functionality and performance with other DICOMweb implementations
• Brainstorm about non-standard extensions of utility to the community
  • search by reference (e.g. find all segmentations of this image)
  • extract cohorts by tag (e.g. find all CTs with .5mm pixel spacing)
  • represent workflow state (e.g. give list of all studies that do not have manual segmentations)
• Work on improving performance

Approach and Plan

Progress and Next Steps

Hans tested the server successfully, in order to try out DICOMWeb and get a feel for what’s possible and how to integrate with other tools. Here’s one quick guide on how to possibly reproduce:

• Hans used CouchDB via Docker (Steve & Emel recommended running PouchDB via NodeJS):

  docker run --rm -d --name couchdb -p 5984:5984 apache/couchdb
  

  (This is for one-off experiments; one should really mount a DB directory.)

• The dicomweb-server-js README contains instructions on how to start the server and index DICOM files via the Python dicomweb_client.

• Subsequently, one may test the Rest API directly (e.g. via httpie on the default port 5985), look at the underlying CouchDB (UI at http://localhost:5984/_utils), or try the OHIFViewer.

  • test_ohifviewer.html is a standalone file with the necessary JS snippet to configure an OHIFViewer. Adapt the URLs to your server if necessary.

  • Use a HTTP server to serve that file and show it in a browser:

    python -m http.server 9000
    xdg-open http://localhost:9000/test_ohifviewer.html
    

Performance Improvements:

• Extended the dicomweb codebase with parallel execution for database processes which lets us limit the number of concurrent database calls and manage workload

Illustrations

• dicomweb-server: https://github.com/dcmjs-org/dicomweb-server
• SlicerChronicle: https://github.com/pieper/Chronicle
• SlicerChronicle: https://github.com/pieper/SlicerChronicle
• dcmjs: https://github.com/dcmjs-org/dcmjs