Scribd
Upload
52 views

Under The Esteemed Guidance of Assistant Professor Dept of Computer Science and Engineering

The document proposes a system for classifying thorax diseases from chest X-ray images using a DenseNet architecture to focus on disease-critical regions and capture intrinsic relationships between features. It describes developing a computer-aided disease analysis system to assist diagnosis by classifying images and suppressing irrelevant regions, as existing methods using text data were inefficient and inaccurate. The proposed system aims to achieve higher accuracy using deep learning techniques by processing chest images to classify thorax diseases.

Uploaded by

Keerthana Ramesh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
52 views

Under The Esteemed Guidance of Assistant Professor Dept of Computer Science and Engineering

The document proposes a system for classifying thorax diseases from chest X-ray images using a DenseNet architecture to focus on disease-critical regions and capture intrinsic relationships between features. It describes developing a computer-aided disease analysis system to assist diagnosis by classifying images and suppressing irrelevant regions, as existing methods using text data were inefficient and inaccurate. The proposed system aims to achieve higher accuracy using deep learning techniques by processing chest images to classify thorax diseases.

Uploaded by

Keerthana Ramesh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 32

DISCRIMINATIVE FEATURE LEARNING FOR THORAX

DISEASE CLASSIFICATION IN
CHEST X-RAY IMAGES

ANNAMACHARYA INSTITUTE OF TECHNOLOGY AND SCIENCES


(Affiliated to J.N.T.U.A., Anantapur, Approved by A.I.C.T.E, New Delhi)
Utukur (P), C.K.Dinne (V&M), Kadapa-516003
ANDHRA PRADESH.

Under the Esteemed Guidance of


Submitted by
Mr. S.MOHAMMAD ALI M.Tech,
1. G. Mahesh (18HM1A0522) Assistant Professor
2. P. Sai Keerthana(18HM1A0550) Dept of Computer Science and Engineering
3.A.Leena(18HM1A0503)
4.M.Humera Khan(18HM1A0542)
5.A.Charan Kumar Reddy(18HM1A0504)
OBJECTIVE

The main objective of this project is to create an effective system for


classifying Thorax diseases using chest X-ray images using dense
net architecture.
ABSTRACT

• Here we focus on the thorax disease classification problem in chest X-ray


(CXR) images.
• Automatically focus on the disease-critical regions.
• Adaptively capture the intrinsic relationships among different disease
features .
• We used transfer learning technique, named Dense net, to achieve our target
prediction .
• Experiments conducted on this disease data set demonstrate the effectiveness
of the proposed methods.
INTRODUCTION

• Developing a stable and robust computer-aided disease analysis system is


critical to assist disease diagnosis and treatment .
• For the thorax disease classification problem, such a system is necessary to
focus on the potential lesion areas and suppress the noise introduced by the
irrelevant regions.
• Chest X-ray image classification suffers from a large amount of disease-
irrelevant regions.
EXISTING METHOD

•In existing system, a machine learning algorithm was used to detect


the Thorax disease.
•Existing system uses text or categorical data to detect the Thorax
diseases.
• As the input data in text or categorical it is inefficient to predict the
disease and it does not give accurate classification results.
LIMITATIONS

• More Computation memory.


• Time Consuming.
• Difficult to handle.
• Produce rough solutions.
• Based on low - level features.
PROPOSED SYSTEM

•In proposed system we introduce the deep learning techniques which is


used for detecting the disease.
• Here we working with chest images, for classifying the Thorax disease
using the Dense net Algorithm.
•By using this process we get more accuracy and perfect results.
ADVANTAGES

• High accuracy.
• Time Saving.
• Computationally inexpensive.
• Low complexity .
• Easy to implement.
SYSTEM REQUIREMENTS
Hardware Configuration:

• Processor : I3/Intel Processor


• Hard Disk : 160GB(min)
• RAM : 8Gb(min)
Software Configuration:

• Operating System : Windows 7/8/10/11


• Server side Script : HTML, CSS & JS
• IDE : PyCharm
• Technology : Python 3.6+
ARCHITECTURE
MODULES

1. User:

1.1 Registration:
If the user is new to the application. He/ She needs to enter the valid details to
register with the website.

1.2 Login:
User should provide valid credentials to login into the application.

1.3 Prediction:
User needs to upload the image to see whether the patient is having thorax
disease or not.
MODULES

2. System

2.1 Create Dataset:

The dataset containing images of the Retinal Lesion segmentation images with the identified vessel

veins to be classified is split into training and testing dataset with the test size of 20-10%.

2.2 Pre-processing:

Resizing and reshaping the images into appropriate format to train our model.

2.3 Training:

Pre-processed training dataset is used to train our model using Encoder and Decoder architecture.

2.4 Prediction:

The results of our model are displayed on the front end.


UML DIAGRAMS
USE CASE DIAGRAM
CLASS DIAGRAM
SEQUENCE DIAGRAM
COLLABORATION DIAGRAM
DEPLOYMENT DIAGRAM
ACTIVITY DIAGRAM
ER DIAGRAM
ALGORITHM

Dense Net:
• In a traditional feed-forward Convolutional Neural Network (CNN), each convolutional
layer except the first one (which takes in the input), receives the output of the previous
convolutional layer and produces an output feature map that is then passed on to the next
convolutional layer.
• Therefore, for 'L' layers, there are 'L' direct connections.
• However, as the number of layers in the CNN increase, i.e. as they get deeper, the
'vanishing gradient' problem arises.
• This means that as the path for information from the input to the output layers increases, it
can cause certain information to 'vanish' or get lost which reduces the ability of the
network to train effectively.
•DenseNet resolve this problem by modifying the standard CNN architecture and simplifying
the connectivity pattern between layers.

•In a DenseNet architecture, each layer is connected directly with every other layer,

•Hence it is called Densely Connected Convolutional Network.

•For 'L' layers, there are L(L+1)/2 direct connections.


SCREEN SHOTS
SCREEN SHOTS
SCREEN SHOTS
SCREEN SHOTS
SCREEN SHOTS
SCREEN SHOTS
SCREEN SHOTS
SCREEN SHOTS
TEST CASES
TC. No Test case Input Expected Actual Status
Result Result
1 User Enter all fields User gets Registration is Pass
Registration registered successful
2 User If user miss any User not Registration is Pass
Registration field registered unsuccessful

3 User login Give email & User homepage is User homepage Pass
password opened has been
opened
4 User login Give email & User homepage Invalid details pass
wrong password should not opened

5 Upload file Select file to Upload to Uploaded pass


upload database successfully
6 Upload Select file that had Upload file Upload pass
duplicate file already uploaded unsuccessful

You might also like