africanbiotech / pulstrat

A Web Application that predicts the presence of Pneumonia from X-Ray Images using a DenseNet Neural Network model. Tech Stack: React as frontend , Python Flask as backend and Mongo DB.

Pulstrat - A Pneumonia Detecting Platform

A Pneumonia Detecting Web Application Built Using Python Flask(Backend server), React Js(Frontend UI), Tailwind CSS, Mongo DB(Database) & SendGrid

Features

• Upload service for x-ray images
• Instant delivery of results
• User Authentication
• Dashboard to display user's history of scans
• Send reports to registered email
• Caching system logic embedded to obtain history of scans quickly
• Downloadable reports in pdf format

System Architecture

Replication & Usage

Clone the repository

git clone https://github.com/xosteve26/PULSTRAT.git

Creating Environmental Variables

Create a .env file in then frontend & backend folder and add the following

Backend

MONGO_URI = Your Mongo DB ATLAS URI [ Mongo DB connection string ]
SECRET_KEY = Your Flask Secret Key
SENDGRID_API_KEY = Your SendGrid API key [ For email functionality ] [Optional]

Frontend

Frontend env consists of the base url, which is the endpoint of the flask server #For local environment

REACT_APP_BASE_URL=http://localhost:5000 

Install Dependencies (Frontend)

cd frontend
npm install 

Install Dependencies (Backend)

cd backend
pip install -r requirements.txt

Run

# Run frontend (:3000) 
cd frontend
npm run start

# Run backend (:5000)
cd backend
python app.py

Test Dataset

This test dataset consists of a total of 624 images, out of which 234 images belong to the NORMAL class and the remaining 390 belong to the PNEUMONIA class. Kindly download the dataset through this link and feel free to test it out.

Build & Deploy

# Create frontend production build [serve static build]
cd frontend
npm run build
serve -s build