Diagram of system components

digraph {
   node [fixedsize=true width=2.0 height=0.75 fontsize=10 margin="0.0,0.0"];

   // Define the things on the server
   subgraph cluster_server {

      label="OpenREM server";
      tooltip="The server";
      fontname="Helvetica-bold";
      fontsize=16;
      style=filled;
      color=lightgrey;
      node [style=filled color=white];

      // Define the nodes for the data storage, display and retrieval
      webserver [label="Web server¹" fontname="Helvetica" tooltip="Serve web pages to the user" shape="box"];
      python_django [label="OpenREM\nDjango app" fontname="Helvetica" tooltip="Python web framework" shape="box"];
      database [label="Database³" fontname="Helvetica" tooltip="Relational database management system" shape="parallelogram"];
      rabbitmq [label="RabbitMQ\nmessage broker" fontname="Helvetica" tooltip="Message broker" shape="box"];
      celery [label="Celery\ntask queue" fontname="Helvetica" tooltip="Asynchronous task queue" shape="hexagon"];
      skin_dose_map_data [label="Skin dose map\ndata calculation,\nstorage, retrieval" fontname="Helvetica" tooltip="Calculate, store and retrieve skin dose map data" shape="parallelogram"];
      server_media_folder [label="Server file storage\n(Media Home folder)" fontname="Helvetica" tooltip="File storage on the server" shape="parallelogram"];
      data_export [label="Data export to\nlocal file system" fontname="Helvetica" tooltip="Files are made available to the user via a web page URL" shape="box"];

      // Define the links between the data storage, display and retrieval
      webserver -> python_django [dir=both];
      python_django -> database [dir=both label="via psycopg2\nPython adapter" fontsize=8 fontname="Courier"];
      python_django -> rabbitmq;
      rabbitmq -> celery;
      celery -> skin_dose_map_data;
      celery -> data_export;
      skin_dose_map_data -> server_media_folder [dir=both];
      skin_dose_map_data -> python_django [style=dotted dir=both];
      data_export -> server_media_folder;
      data_export -> python_django [style=dotted dir=both];

      // Define the nodes for the DICOM store, database population and skin dose map calculation
      conquest [label="DICOM StoreSCP²" fontname="Helvetica" tooltip="Conquest, acting as a DICOM storage SCP" shape="box"];
      conquest_script [shape=diamond label="Does the\nobject contain\nuseful data?" fontname="Helvetica" tooltip="Process the rules in dicom.ini"];
      populate_database [label="Extract information from\nthe DICOM object to the\nOpenREM database" fontname="Helvetica", tooltip="Extract data using OpenREM's python scripts" shape="box"];
      delete_object [label="Delete the DICOM object\nfrom the DICOM store" fontname="Helvetica" tooltip="Delete the DICOM object from the local store SCP" shape="box"];
      calc_skin_dose_map [shape=diamond label="Calculate\nskin dose\nmap?" fontname="Helvetica" tooltip="Calculate the skin dose map?"];
      blank_node_1 [shape=none style=invisible];

      // Define the links between the DICOM store, database population and skin dose map calculation
      conquest_script -> populate_database [label="Yes" fontcolor=darkgreen fontsize=8 fontname="Courier"];
      populate_database -> delete_object;
      conquest_script -> delete_object [label="No" fontcolor=red fontsize=8 fontname="Courier"];
      conquest -> conquest_script;
      populate_database -> calc_skin_dose_map;

      // Define the links between the two groups
      python_django -> populate_database [dir=back]
      calc_skin_dose_map -> celery [style=dotted label="Yes" fontcolor=darkgreen fontsize=8 fontname="Courier"]

      // Force certain nodes to be on the same level so that the diagram looks good (hopefully)
      {rank=same; webserver conquest};
      {rank=same; celery calc_skin_dose_map };
      {rank=same; python_django->blank_node_1->delete_object->populate_database [style=invis]; rankdir=LR;}
   }

   // Define the web browser, modality and pacs nodes
   web_browser [label="Client\nweb browser" fontname="Helvetica" tooltip="The user's web browser" shape="box" style=rounded];
   modality [label="X-ray imaging\nmodality" fontname="Helvetica" tooltip="Data send from an x-ray imaging modality" shape="parallelogram"];
   pacs [label="PACS" fontname="Helvetica" tooltip="A Picture Archiving and Communication System" shape="parallelogram"];
   blank_node_2 [label="" shape=none];

   // Define the links that the browser, modality and pacs have
   web_browser -> webserver [dir=both label="via user-requests\land Ajax\l" fontsize=8 fontname="Courier" tooltip="Ajax used to retrieve chart data"];
   modality -> conquest [label="via modality\lconfiguration\l" fontsize=8 fontname="Courier"];
   pacs -> conquest [label="via OpenREM\lquery-retrieve\l" fontsize=8 fontname="Courier"];

   // Force the web browser, blank node, modality and pacs to be on the same level in a specific order
   {rank=same; web_browser->blank_node_2->modality->pacs [style=invis]; rankdir=LR;};
}

Alternatives

1: Web servers

The recommended web server for Windows is Microsoft IIS - see Running OpenREM on Windows with IIS for details. This has replaced the recommendation to use Apache due to difficulties in obtaining the required binary files, as described in the Web servers section of the installation document.

The recommended web server for Linux is Gunicorn with NGINX - see Running OpenREM on Linux with Gunicorn and NGINX for details.

Alternatively, a built-in web server is included that will suffice for testing purposes and getting started.

2: DICOM Store node

Any DICOM Store can be used, as long as it can be used to call the OpenREM import script. A built-in store is available, but not recommended for production use. See DICOM Network Configuration for more details. Orthanc or Conquest are the recommended DICOM Store services to use; see Install a DICOM Store service for installation notes and the Third-party DICOM Stores section for configuration help.

3: Database

PostgreSQL is the recommended database to use with OpenREM. It is the only database that OpenREM will calculate median values for charts with. Other databases can be used with varying capabilities; see the Django documentation for more details. For testing only, the built-in SQLite3 database can be used, but this is not suitable for later migration to a production database.