syedissambukhari/Pneumonia_Detection_using_CNN

Pneumonia Detection using CNN

Pneumonia Detection using CNN
Explore the docs »

View Demo · Report Bug

Table of Contents

• Table of Contents
• About The Project
  • Steps involved
  • Tools used
• Getting Started
  • Installation
  • Data Analysis and Model Building
  • Application Development
• Usage
• Contact

About The Project

This project uses Chest X-Ray dataset from kaggle. The dataset is organized into 3 folders (train, val & test) and contains subfolders for each image category (Pneumonia/Normal). There are 5,863 X-Ray images (JPEG) and 2 categories (Pneumonia/Normal). These images are used for building an CNN model using keras to detect pneumonia in patients based on X-ray reports. After model building the best performing model is used to build a flask base web application.

Steps involved

• Reading data from source and preprocessing it using OpenCV.
• Performing data preprocessing before feeding the data to the model.
• Building sequential model architecture using keras.
• Using data augmentating to prevent the model from overfitting during training phase.
• Creating web application using flask to detect Pneumonia upon receiving the preprocessed X-ray.

Tools used

Following are the tools/frameworks used in developing the application:

• CSS
• Python
• Flask
• HTML
• OpenCV

Getting Started

Installation

For proper execution of application firstly create an environment, then to install prerequisite libraries execute below command in terminal.

pip install -r requirements.txt

Data Analysis and Model Building

Refer Pneumonia_Detection_CNN.ipynb to find details regarding data analysis and model building.

Since this is a classification problem, to check model performance Confusion Matrix and Classification Report are used.

Confusion Matrix

Application Development

After model development, the web application is developed using flask which is a python based web-framework. For source code refer app.py.

Below are few snapshots of application in use:

• Step 1 Uploading X-Ray

  

• Step 2 Displaying Result

  

Usage

To run this application firstly execute python app.py, after which the flask built-in server would start hosting the application at localhost i.e. http://127.0.0.1:5000/