Skip to content

A model for Pnuemonia disease prediction using the X-Ray images and applying basic image feature extraction, machine learning models and web deployment using Flask

Notifications You must be signed in to change notification settings

Gaurav0502/pneumonia-detection-system

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

41 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Problem Statement

Medical Image Analysis is used to process medical images like X-rays(here),CT Scans etc. Here, we will classify whether or not that person suffers from a disease or not. Moreover, it can also be used for plant disease prediction as well.

Dataset

The dataset used is the Chest X-Ray Images (Pneumonia)(https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia) from Kaggle.

Task : Classification

The 2 class labels are:

1. Pneumonia: Given X-Ray is of a person suffering from Pneumonia.

2. Normal: Given X-Ray is of a normal person.

Model(s) Used

Various models were trained with the grayscale X-Rays like :

  1. Binary Logistic Regression.
  2. K- Nearest Neighbours (KNN).
  3. Support Vector Machines (SVM).
  4. Multilayer Perceptron (MLP).
  5. Neural Network.

Binary Logistic Regression is a Supervised Learning model in which the sigmoid function is used to predict a categorical dependent variable (here, X-Ray type) based on a given set of independent variables.

K-Nearest Neighbours is a Supervised Learing model in which is used for both classification and regression tasks. Here, we use KNN for classifying the type of Lung X-Ray. This algorithm classifies test values based k-nearest train values usually based on Euclidean distance metric.

Support Vector Machines is a Supervised Learning model which can be used for both classification and regression problems. The main idea behind this algorithm is estimate the best possible decision boundary to segreggate a n-dimensional space called hyperplanes. For a new set of independent variables it classifies the dependent variable based on the distance boundary or hyperplane.

Multilayer Perceptron is a Supervised Learning model which is majorly used for classification tasks. It is type feedforward Artificial Neural Network that generates a set of outputs from a set of inputs. Diagrammatically, it is a directed graph with the outermost layers as the input and output layers and the intermediate layers referred to as inner layers. The model is trained using back propagation technique.

Blog Link

The project is documented as a medium article.

Future Work

To apply much better image processing techniques and training deep learning models.

Contributors

Gaurav Pendharkar https://github.com/Gaurav0502.
Prasham Titiya https://github.com/prasham1515.

About

A model for Pnuemonia disease prediction using the X-Ray images and applying basic image feature extraction, machine learning models and web deployment using Flask

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages