Diagnosing Pneumonia
from Chest X-Rays using
Neural Networks
 National College Of Ireland
 Module : Advance Data Mining
 Team - C
 Ashish Soni: x18136664 , Tushar S. Dalvi |
x18134301, Shantanu Deshpande |
x18125514, Yash Iyengar | x18124739
Contents
Introduction
Research Question
Data Gathering
Data Pre-processing
Results
Introduction
Radiology is a branch of medicine where the disease
diagnosed by examining X-ray Images.
To reduce the human eye error in diagnosing the disease
computer aided system has evolve for better diagnosis.
Machine learning techniques has shown remarkable
results in the recent years.
In this project we are trying to diagnose the Pneumonia
from chest X-ray using Machine Learning techniques.
Research
Question
 Can the recall be improved for Chest X-ray
Pneumonia Detection using CNN based VGG19
model as compared to the state of the art
technique?
Data
Gathering
Data for this project is extracted from
Kaggle.
It contains 1341 normal lung images of X-
rays and 3875 pneumonia infected images.
Images are in grayscale format and of
varying sizes.
Data
Preprocessing
Rescaling of the images is done by dividing each
image by 255.
Images are resized to 128 by 128 pixels.
Rotation of images performed by 20 degrees
horizontal rotation.
Width and height shifted by 0.2 fractions of the
total width and with the input flipped horizontally.
Results
Precision = TP / (TP + FP)
Recall = TP / (TP + FN)
Acknowledgement
We would like to express our sincere gratitude
to Mr. Noel Cosgrave for his mentorship and
guidance.

Pneumonia detection using cnn

  • 1.
    Diagnosing Pneumonia from ChestX-Rays using Neural Networks  National College Of Ireland  Module : Advance Data Mining  Team - C  Ashish Soni: x18136664 , Tushar S. Dalvi | x18134301, Shantanu Deshpande | x18125514, Yash Iyengar | x18124739
  • 2.
  • 3.
    Introduction Radiology is abranch of medicine where the disease diagnosed by examining X-ray Images. To reduce the human eye error in diagnosing the disease computer aided system has evolve for better diagnosis. Machine learning techniques has shown remarkable results in the recent years. In this project we are trying to diagnose the Pneumonia from chest X-ray using Machine Learning techniques.
  • 4.
    Research Question  Can therecall be improved for Chest X-ray Pneumonia Detection using CNN based VGG19 model as compared to the state of the art technique?
  • 5.
    Data Gathering Data for thisproject is extracted from Kaggle. It contains 1341 normal lung images of X- rays and 3875 pneumonia infected images. Images are in grayscale format and of varying sizes.
  • 6.
    Data Preprocessing Rescaling of theimages is done by dividing each image by 255. Images are resized to 128 by 128 pixels. Rotation of images performed by 20 degrees horizontal rotation. Width and height shifted by 0.2 fractions of the total width and with the input flipped horizontally.
  • 7.
    Results Precision = TP/ (TP + FP) Recall = TP / (TP + FN)
  • 8.
    Acknowledgement We would liketo express our sincere gratitude to Mr. Noel Cosgrave for his mentorship and guidance.