Skip to content

Latest commit

 

History

History
52 lines (45 loc) · 3.37 KB

components.md

File metadata and controls

52 lines (45 loc) · 3.37 KB
Raw

Components:

Phase 1 Components

Classification Server

Main file: ServerMain.py

  • Runs on the cloud computer to do model training and classification.
  • Run an event loop that listens on a encrypted TCP socket and port for requests.
  • An initial request establishes which processing model to use, the only current supported is 'rtAtten'.
  • Once the model is selected it will forward command requests to the model file's handleMessage() function.
  • See files rtfMRI/RtfMRIServer.py for the top level request handler, and rtAtten/RtAttenModel.py for the model specific message handling functions.

Client

Main file: ClientMain.py

  • Runs on the local computer (in the control room)
  • Accepts a parameter -m indicating which model to use (such as 'rtAtten')
  • The model is the framework for the particular experiment, such as how to initialize the experiment, the blocks of TRs to run etc.
  • The client connects to the server and runs the model making processing requests to the server as needed
  • The only model currently supported is 'rtAtten'
  • See files rtfMRI/RtfMRIClient.py for top level handling and rtAtten/RtAttenClient.py for model specific handling.

Phase II Components - Web-Based Project Interface

WebServer

Main file: WebMain.py

  • A web server running on the cloud computer that runs both the ClientMain.py and ServerMain.py (components described above) when requested to start a run
  • Provides a browser based control panel for starting and stopping runs, viewing graphs of results, changing settings and registration steps.
  • Accepts https connections from the experimenter's to provide a control panel
  • Uses browser web-socket connection to provide logging, graphing and VNC display of registration results to the experimenter's browser
  • Accepts web-socket connections from the FileWatchServer running on the local computer to read Dicom files as the become available from the scanner
  • VNC server run with websockify as front-end to accept websocket connections for display in client browser. See scripts/run-vnc.sh and websockify conda environment
  • See webInterface/ directory, webInterface/WebServer.py
  • See webInterface/rtAtten/web for browser code and webInterface/rtAtten/registration for registration scripts
  • See websockify.yml for installing websockify conda environment

FileWatchServer

Main file: FileWatchServer.py

  • Runs on the local computer (in the control room) and listens for file requests, such as to watch for a Dicom file
  • Retrieves the file being watched for when it is available and returns it to the webServer via web-sockets
  • Accepts data files to be written locally from the webServer, primarily for supplying classification results back to the local computer for subject feedback display handled by PsychToolbox or PsychoPy
  • See file webInterface/webSocketFileWatcher.py

Browser code

Main webServer file: webInterface/rtAtten/RtAttenWeb.py

  • Called from WebMain.py and uses webInterface/webServer.py
  • Browser
    • JavaScript code in webInterface/rtAtten/web/src
    • Uses React framework for handling events and rendering
    • Uses websockify in front of VNC for showing registration results in browser
  • Registration scripts in webInterface/rtAtten/registration
  • Connect to main control window at URL https://<cloud_ip_addr>:8888
  • Connect to subject feedback window at URL https://<cloud_ip_addr>:8888/feedback