Note: This project is just a proof of concept prototype and should not be used for Medical use.
Blaze is a webapp designed to predict the probabilities of different pathologies visible under a human chest xray. It aims to co-pilot and assist medical professionals to diagnose patients accurately and efficiently. This project was made as a submission for a competition and is based on the torchxrayvision library.
Firstly, you need to make sure you have installed and started up the server. After you start it you get one of two URL's, which will look like the following: http://0.0.0.0:13520
or something similar to http://90eu-285-11-20-34.ngrok.io
. After that, you can simply visit the url to use the webapp. Note that the first type of URL is only accessible from the host device.
Note: As of now the downloading without python is not supported but will be in the very near future
To get a pre-compiled version, you can visit the releases section and download the latest version. The rest of the instructions will be given in the following sections. For MacOS, you will need to download as a python project, this is due to Apple only allowing the execution of apps compiled on Apple Hardware.
Firstly, you need to install python, you can get this from the python website. When running the installer, please check the "Add to PATH" option, otherwise python will not work. You may also install python from a commandline package manager (if you are make sure to install pip as well)
If you have the Git Version Control software installed then you can you can type the following in your terminal:
git clone https://github.com/SystematicError/Chest-Xray-AI
If you do not have git installed or wish to do it in a more graphical way, then you can also get a zip file containing the project from here; afterwards extract this zip file.
Now, open your terminal, and navigate to the extracted directory using command prompt or your appropriate terminal (here you can find a guide for windows or go here for a Linux/MacOS guide)
When you have navigated to the right directory, type the following; it may take a minute to install depending on your internet speed:
pip install -r requirements.txt
When starting the server for the first time, it will download the AI Models and generate some cache. There are 2 ways you can start the server:
To open it graphically, you can open it via your file explorer. Navigate to the project folder and double click the main.py
file (it could also be main.exe
or simply main
), make sure that when you double click that the program it opens with python and not some text editor. A terminal window should pop up if everything is done correct. If the window opens and quickly closes that means there is some error, try running the server via commandline to find the root of the issue.
Firstly open your terminal of choice and navigate to the project directory, instructions on how to do so is given in the "Downloading as python project" section.
If you installed as a python project, then run the following:
python main.py
Otherwise, run the these commands instead:
main.exe
./main
If all goes well you should see something like this:
[xray] Importing libraries
[xray] Loading model
[server] Running locally at http://0.0.0.0:13520 (http://localhost:13520)
[tunnel] Created tunnel at http://90eu-285-11-20-34.ngrok.io
The first URL, http://0.0.0.0:13520
is your local URL, you can only visit this website from the computer the server is running from. This URL will always be the same. If your browser doesn't load the page try loading the URL in the brackets, http://localhost:13520
.
The next URL, http://90eu-285-11-20-34.ngrok.io
is your public URL, this can be visited from any device as long as the server is running. This URL will change whenever you restart the server.
Note: You can press Ctrl+C
in the terminal to stop the server
If you face an error while launching the server, try re-doing the command, if there is still an issue please create a issue on our issue tracker.