Skip to content
/ metacat Public

Custom medical imaging metadata catalogue including classification based on DICOM metadata

Notifications You must be signed in to change notification settings

SMI/metacat

Repository files navigation

Metacat

This repository contains documents and scripts for the creation of a custom medical imaging metadata catalogue. This includes the following processes:

  • Statistical metadata generation at different stages of the SMI pipeline (raw, staging, live)
  • Derived metadata generation from DICOM metadata (e.g., DICOM-metadata based classification by body part)
  • Import of DICOM standard metadata
  • Creation of visualisations based on generated metadata via a web interface

Repo contents

Directory Contents
docs Metadata and table schemas, overview of the data architecture.
metadata_collection Scripts for the generation and collection of statistical metadata and storage in a MongoDB database.
metadata_studies Derived metadata studies and experiments, including scripts for extracting metadata for analysis.
catalogue_ui Implementation of Flask app as the catalogue UI.
modules Shared document and database manipulation.
test Test deployment of all components in a Docker environment.

Deploy with Docker

If you want to change configuration (i.e., front-end serving port, database credentials), source an environment file like config.env before the docker-compose command:

$ source test/config.env
$ docker-compose -f test/docker-compose.yml up

Otherwise, the default values from the compose file will be used.

This will launch the following containers:

  • smi-mariadb: acting as both staging and live database
  • smi-mongodb: acting as raw and metadata database
  • smi-catalogue: host of catalogue front and back end

And will perform the following tasks:

  1. Install requirements.txt and custom modules in the smi-catalogue container.
  2. Make use of the DICOM metadata schema to generate synthetic documents in a dicom database on smi-mongodb.
  3. Make use of the table schema to create tables on smi-mariadb and populate them with data from the dicom MongoDB database. To emulate the processed data on the live system, two databases are populated, data_load2 representing the staging database, and smi representing the live database.
  4. Perform metadata collection tasks (see documentation for more details).
  5. Run body part labelling (see documentation for more details).
  6. Deploy catalogue UI.

You can manually re-run any of these tasks from the smi-catalogue container. For example:

(smi-catalogue) # . /home/metacat/test/config.env
(smi-catalogue) # cd /home/metacat/metadata_collection
(smi-catalogue) # python3 populate_catalogue.py -d dicom -i -l logs/

And you can manually analyse the MongoDB documents. If you used the default configuration, note that the username password are in the config.env file:

(smi-mongodb) # mongosh -u <MONGOUSER> -p
<MONGOPASS>

Similarly for the MariaDB container:

(smi-mariadb) # mariadb -u <MYSQLUSER> -p
<MYSQLPASS>

To stop containers, press CTRL+C and run the following command to clean up:

$ docker-compose -f ./test/docker-compose.yml down

About

Custom medical imaging metadata catalogue including classification based on DICOM metadata

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published