LUNG HISTOPATHOLOGICAL IMAGE

CLASSIFICATION USING TENSORFLOW WITH KERAS

A PROJECT REPORT

Submitted by

ANUGRAHA C (113119UG07008)

DEEPA V (113119UG07016)

GRACELIN C (113119UG07027)

RESHMA K (113119UG07075)

in partial fulfillment for the award of the degree

of
BACHELOR OF TECHNOLOGY

IN

INFORMATION TECHNOLOGY

VEL TECH MULTI TECH Dr. RANGARAJAN Dr. SAKUNTHALA

ENGINEERING COLLEGE,
ALAMATHI ROAD, AVADI, CHENNAI-62

ANNA UNIVERSITY, CHENNAI 600 025

JUNE 2022
1
 BONAFIDE CERTIFICATE

Certified that this project report of title “LUNG HISTOPATHOLOGICAL IMAGE

CLASSIFICATION USING TENSORFLOW WITH KERAS” is the bonafide work of
ANUGRAHA C (113119UG07008), DEEPA V (113119UG07016), GRACELIN C
(113119UG07027) and RESHMA K (113119UG07075) who carried out the project work
under my supervision. Certificate further, that to the best of my knowledge the work reported
here is does not form any other project report or dissertation on the basis of which a degree or
award was conferred on an earlier occasion on this or any other candidate.

SIGNATURE SIGNATURE

HEAD OF THE DEPARTMENT INTERNAL GUIDE

Mr. R. PRABU, M. Tech. Dr. M. RAJESH KHANNA, PhD.
HEAD OF THE DEPARTMENT ASSOCIATE PROFESSOR
Department of Information Department of Information
Technology Technology
Vel Tech Multi Tech Dr. Rangarajan Vel Tech Multi Tech Dr.Rangarajan
Dr. Sakunthala Engineering College, Dr.Sakunthala Engineering College,
Avadi, Chennai-600 062 Avadi, Chennai-600 062

2
 CERTIFICATE FOR EVALUATION

College Name : VEL TECH MULTI TECH Dr. RANGARAJAN Dr. SAKUNTHALA
ENGINEERING COLLEGE.
Branch : INFORMATION TECHNOLOGY
Semester : VII

S.NO NAME OF THE TITLE OF THE NAME OF THE

STUDENTS WHO PROJECT SUPERVISOR
HAS DONE THE
PROJECT
1. ANUGRAHA C LUNG Dr. M. RAJESH
[113119UG07008] HISTOPATHOLOGICAL KHANNA, PhD.
IMAGE
2. DEEPA V CLASSIFICATION ASSOCIATE
[113119UG07016] USING TENSORFLOW PROFESSOR,
WITH KERAS Department of
3. GRACELIN C Information
[113119UG07027] Technology

4. RESHMA K
[113119UG07075]

The report of this project work submitted by the above students in partial
fulfilment for the award of Degree of Bachelor of Technology in Information
Technology of Anna University was evaluated and confirmed to be the report of
work done by the above student. This project report was submitted for the viva-
voice held on ____________ at Vel Tech Multi Tech Dr.Rangarajan
Dr.Sakunthala Engineering College.

INTERNAL EXAMINER EXTERNAL EXAMINER

3
 ACKNOWLEDGEMENT

We wish to express our sincere thanks to almighty and the people who extended their
help during the course of our work.
We are greatly and profoundly thankful to our honourable Chairman, Col. Prof.
Vel. Shri Dr.Rangarajan B.E.(ELEC), B.E.(MECH), M.S.(AUTO)., D.Sc., &
Vice Chairman, Dr. Sakunthala Rangarajan M.B.B.S., for facilitating us with
this opportunity.
We take this opportunity to extend our gratefulness to our respectable Chairperson
& Managing Trustee Smt. Rangarajan Mahalakshmi Kishore B.E., M.B.A.,
for her continuous encouragement.
Our special thanks to our cherishable Vice- President Mr.K.V.D. Kishore Kumar
B.E., M.B.A., for his attention towards students’ community.
We also record our sincere thanks to our honorable Principal, Dr.V. Rajamani M.E.,
Ph.D., for his kind support to take up this project and complete it successfully.
We would like to express our special thanks to our Head of the Department,
Mr.M.Prabhu M.Tech, Department of Information Technology, our internal
guide Dr. M. Rajesh Khanna PhD, and our department project coordinator Dr.
M. Rajesh Khanna PhD, for their moral support by taking keen interest on our
project work and guided us all along, till the completion of our project work and
also by providing with all the necessary information required for developing a
good system with successful completion of the same.
Further, the acknowledgement would be incomplete if we would not mention a
word of thanks to our most beloved Parents for their continuous support and
encouragement all the way through the course that has led us to pursue the degree
and confidently complete the project work.

4
 ABSTRACT

The human respiratory system's most important organs are the lungs. One of the most
devastating and ubiquitous diseases, cancer results in a significant amount of death annually.
Lung cancer is the most dominant and has the largest mortality of all the distinct types of
cancer. Histopathology tissue is the section of microscopic lung tissue with affected diseases
and it is initial step to identify the cancer. The gold standard for qualitative and clinical lung
cancer staging is histopathological evaluation. However, it is quite tedious for doctors to
analyse hundreds of histopathological images, especially for those with less experience. As a
result, objective pathological diagnosis outcomes can successfully lead doctors in identifying
the best treatments, increasing patient survival rates. Traditionally lung cancer has been only
predicted. Our main objective of this project is to classify the lung tissue using
histopathological image and implement CNN algorithm (LeNet and VGG-16) for better
analysis of lung cancer with the help of TensorFlow with Keras.

5
 TABLE OF CONTENTS

CHAPTER TITLE PAGE NO

ABSTRACT

LIST OF FIGURES
1. INTRODUCTION

1.1. DEFINITION

1.2. OBJECTIVE

1.3. SCOPE OF THE PROJECT

2. LITERATURE SURVEY

2.1. PAPER 1
2.1.1. ABSTRACT
2.1.2. ADVANTAGES & DISADVANTAGES
2.2. PAPER 2
2.2.1. ABSTRACT
2.2.2. ADVANTAGES & DISADVANTAGES
2.3. PAPER 3
2.3.1. ABSTARCT
2.3.2. ADVANTAGES & DISADVANTAGES
2.4. PAPER 4
2.4.1. ABSTRACT
2.4.2. ADVANTAGES & DISADVANTAGES
2.5. PAPER 5
2.5.1. ABSTRACT
2.5.2. ADVANTAGES & DISADVANTAGES
6
3. SYSTEM DESIGN AND IMPLEMENTATION

3.1. SYSTEM REQUIREMENTS

3.1.1. HARDWARE REQUIREMENTS
3.1.2. SOFTWARE REQUIREMENTS
3.2. EXISTING SYSTEM
3.2.1. DISADVANTAGES OF EXISTING
SYSTEM
3.3. PROPOSED SYSTEM
3.3.1 ADVANTAGES OF PROPOSED
SYSTEM
3.4. SYSTEM ARCHITECTURE

4. MODULES DESCRIPTION

4.1. OVERVIEW OF THE PROJECT

4.2. MODULES
4.2.1 DATASET COLLECTION & PRE-
PROCESSING
4.2.2 TRAINING AND TESTING THE DATA
4.2.3 CLASSIFICATION IDENTIFICATION
4.2.4 DEPLOYMENT
4.3. UML DIAGRAMS
4.3.1 USE CASE DIAGRAM
4.3.2 SEQUENCE DIAGRAM
4.3.3 CLASS DIAGRAM
4.3.4 ER DIAGRAM
4.3.5 ACTIVITY DIAGRAM

7
 4.3.6 DATAFLOW DIAGRAM

5. CONCLUSION AND FUTURE WORK

5.1 CONCLUSION
5.2 FUTURE WORK

8
 LIST OF FIGURES

FIGURE NO NAME PAGE NO

1 DFD FLOW DIAGRAM

2 SYSTEM ARCHITECTURE
3 USE CASE DIAGRAM
4 CLASS DIAGRAM
5 SEQUENCE DIAGRAM
6 ER DIAGRAM
7 DATA FLOW DIAGRAM

9
 CHAPTER 1
INTRODUCTION

10
 INTRODUCTION

1.1 INTRODUCTION:
In your chest are two sponge-like organs called your lungs. The right lung in your
body is divided into three lobes. Two lobes make up your left lung. Because the heart occupies
more space on the left side of the body, the left lung is smaller. When you inhale, air enters
through your mouth or nose and travels through your trachea to your lungs (windpipe). The
trachea splits into bronchi, which pass into the lungs and further split into smaller bronchi.
These divide into bronchioles, which are smaller branches. The tiny air sacs known as alveoli
are located at the end of the bronchioles. When you inhale air, the alveoli take oxygen into
your blood and eliminate carbon dioxide when you exhale. The major jobs of your lungs are
to take in oxygen and expel carbon dioxide. The cells lining the bronchi and other areas of the
lung, such as the bronchioles or alveoli, are where lung malignancies generally begin. An
depiction of the lungs and their surroundings. The pleura, a slender layer of lining, encircles
the lungs. Your pleura shields your lungs and aids in their movement back and forth against
the chest wall as you breathe. The diaphragm, a slender, dome-shaped muscle, divides the
chest from the belly under the lungs. The diaphragm contracts and expands while you breathe,
propelling air into and out of the lungs.

Types of lung cancer

There are 2 main types of lung cancer and they are treated very differently.

Non-small cell lung cancer (NSCLC)

About 80% to 85% of lung cancers are NSCLC. The main subtypes of NSCLC are
adenocarcinoma, squamous cell carcinoma, and large cell carcinoma. These subtypes, which
start from different types of lung cells are grouped together as NSCLC because their treatment
and prognoses (outlook) are often similar.

Adenocarcinoma: Adenocarcinomas start in the cells that would normally secrete substances
such as mucus.

11
This type of lung cancer occurs mainly in people who currently smoke or formerly smoked,
but it is also the most common type of lung cancer seen in people who don't smoke. It is more
common in women than in men, and it is more likely to occur in younger people than other
types of lung cancer. Adenocarcinoma is usually found in the outer parts of the lung and is
more likely to be found before it has spread.

People with a type of adenocarcinoma called adenocarcinoma in situ (previously

called bronchioloalveolar carcinoma) tend to have a better outlook than those with other
types of lung cancer.

Squamous cell carcinoma: Squamous cell carcinomas start in squamous cells, which are flat
cells that line the inside of the airways in the lungs. They are often linked to a history of
smoking and tend to be found in the central part of the lungs, near a main airway (bronchus).

Large cell (undifferentiated) carcinoma: Large cell carcinoma can appear in any part of the
lung. It tends to grow and spread quickly, which can make it harder to treat. A subtype of large
cell carcinoma, known as large cell neuroendocrine carcinoma, is a fast-growing cancer that
is very similar to small cell lung cancer.

Other subtypes: A few other subtypes of NSCLC, such as adenosquamous carcinoma and
sarcomatoid carcinoma, are much less common.

Small cell lung cancer (SCLC)

About 10% to 15% of all lung cancers are SCLC and it is sometimes called oat
cell cancer. This type of lung cancer tends to grow and spread faster than NSCLC. About 70%
of people with SCLC will have cancer that has already spread at the time they are diagnosed.
Since this cancer grows quickly, it tends to respond well to chemotherapy and radiation
therapy. Unfortunately, for most people, the cancer will return at some point.

Other types of lung tumors

12
 Along with the main types of lung cancer, other tumors can occur in the lungs.

Lung carcinoid tumors: Carcinoid tumors of the lung account for fewer than 5% of lung
tumors. Most of these grow slowly. For more information about these tumors, see Lung
Carcinoid Tumor.

Other lung tumors: Other types of lung cancer such as adenoid cystic carcinomas,
lymphomas, and sarcomas, as well as benign lung tumors such as hamartomas are rare. These
are treated differently from the more common lung cancers and are not discussed here.

Cancers that spread to the lungs: Cancers that start in other organs (such as
the breast, pancreas, kidney, or skin) can sometimes spread (metastasize) to the lungs, but
these are not lung cancers. For example, cancer that starts in the breast and spreads to the lungs
is still breast cancer, not lung cancer. Treatment for metastatic cancer to the lungs is based on
where it started (the primary cancer site).

CANCER DETECTION TECHNIQUES

The primary LC imaging examination techniques are as follows: (1) X-ray

photography, one of the most fundamental lung imaging inspection methods, but the
photography's resolution is inadequate and the examination has blind spots. Chest computed
tomography (CT), in particular CT provides advantages in locating the lesion and finding early
peripheral LC. At the moment, one of the most popular imaging techniques for preoperative
stage and diagnosis of LC. However, one of the limitations of using CT scans is that when
patients are undergoing many scans, it is vital to take into account the effects of the radiation
dose that the procedure produced High sensitivity and specificity are characteristics of
magnetic resonance imaging (MRI). For vertebral and bone metastases, magnetic resonance
imaging (MRI) has a good sensitivity and specificity, however it is not advised for a regular
LC diagnostic. A non-invasive technology called ultrasound is typically more effective than
radiography for examining postoperative pulmonary problems (PPCs). It has grown to be a
useful technique . Although the aforementioned image examination techniques are crucial for

13
LC detection, Each examination's results are solely used as a guide for diagnosis, staging, re-
staging, effectiveness monitoring, and While histopathological examination is the gold
standard for tumour quality and clinical prognostic evaluation of LC.

1.2 OBJECTIVES:

The goal is to develop a deep learning model for Lung Histopathology image
classification by convolutional neural network algorithm for potentially classifying the results
in the form of best accuracy by comparing the CNN architectures.

1.3 SCOPE OF THE PROJECT:

Most of the related works classify the disease with two classifiers. But in this project,
we are going to classify three classifiers. The scope of our research is to split these classes into
adenocarcinoma, benign tissue, and squamous cell carcinoma of the lung. This can be
classified into 3 classes. Besides, extracting handcrafted features from raw images after
different processing is one of the best scopes we have worked on in this research. The objective
of using this method is to get some features which are specifically responsible for lung
histopathology prediction.

14
 CHAPTER 2
LITERATURE SURVEY

15
2.1 Ultrasound Elastography for Lung Disease Assessment
Boran Zhou , Member, IEEE, Xiaofeng Yang , Xiaoming Zhang , Senior Member,
IEEE, Walter J. Curran, and Tian Liu (2020)
Ultrasound elastography (US-E) is a non-invasive, safe, cost-effective and reliable
technique to assess the mechanical properties of soft tissue and provide imaging biomarkers
for pathological processes. Many lung diseases such as acute respiratory distress syndrome,
chronic obstructive pulmonary disease, and interstitial lung disease are associated with
dramatic changes in mechanical properties of lung tissues. Nevertheless, US-E is rarely used
to image the lung because it is filled with air. The large difference in acoustic impedance
between air and lung tissue results in the reflection of the ultrasound wave at the lung surface
and, consequently, the loss of most ultrasound energy. In recent years, there has been an
increasing interest in US-E applications in evaluating lung diseases. This article provides a
comprehensive review of the technological advances of US-E research on lung disease
diagnosis. We introduce the basic principles and major techniques of US-E and provide
information on various applications in lung disease assessment. Finally, the potential
applications of US-E to the diagnosis of COVID-19 pneumonia is discussed.

2.1.2 ADVANTAGES AND DISADAVANTAGES

ADVANTAGES:
• This is a reliable technique to assess the mechanical properties of soft tissue and provide
imaging biomarkers for pathological processes.
• Reduce the risk of spreading the COVID-19 infection among the healthcare personnel
• Reduce exposure to radiation from CT Scans.
DISADAVANTAGES:
• This technique is not commonly used for lung disease diagnosis
• It requires precise longitudinal tracking (hour-to-hour or day-to-day) of an individual
patient’s lung condition to evaluate the efficacy of the management of the disease.

2.2 Using a Noisy U-Net for Detecting Lung Nodule Candidates

Wenkai Huang 1, (MEMBER, IEEE), and Lingkai hu2 (2019)
Chest computed tomography is an important method for detecting lung cancer. Because early
lung cancer nodules are small, making them difficult to detect, the current automatic lung
cancer detection system can easily lead to missed diagnosis when detecting these nodules;
therefore, accurate detection of early lung cancer nodules is crucial for improving the lung
16
cancer cure rate. To reduce the missed diagnosis rate of early nodules in the detection system,
it is necessary to optimize the extraction steps of candidate nodules. Based on the improvement
of U-Net, this paper proposes noisy U-Net (NU-Net), which can enhance the neural network’s
sensitivity to small nodules by adding a special noise to the hidden layers in training. The
neural network is trained using the Lung Nodule Analysis 2016 dataset and the Alibaba
Tianchi Lung Cancer Detection Competition dataset. The comparative experiment between
the U-Net and NU-Net reveals that the proposed algorithm’s sensitivity to small nodules with
diameters of 3–5 mm is greater than the U-Net value.
2.2.1 ADVANTAGES AND DISADAVANTAGES
ADVANTAGES:
• The noisy U-Net can enhance the neural network’s sensitivity to small nodules by
adding a special noise to the hidden layers in training.
• Helps in reducing the missed diagnosis rate of lung cancer diagnosis systems
• It is more sensitive to nodule detection and this difference is more prominent in the
detection of small nodules.

DISADVANTAGES:
• Because the NU-Net uses a regularization method relative to U-Net, it will increase the
difficulty of training
• Since large images needed to be detected in this paper, using a three-dimensional
convolution neural network would have resulted in insufficient memory and an
untrainable network.

2.3 Research on the Auxiliary Classification and Diagnosis of Lung Cancer Subtypes
Based on Histopathological Images
Min li, Xiaojian ma, Chen Chen, Yushuai yuan, Shuailei Zhang, Ziwei yan, Cheng Chen,
Fangfang chen, Yujie bai, Panyun Zhou, Xiaoyi lv and Mingrui ma

Lung cancer (LC) is one of the most serious cancers threatening human
health. Histopathological examination is the gold standard for qualitative and clinical staging
of lung Tumor. However, the process for doctors to examine thousands of histopathological
images is very cumbersome, especially for doctors with less experience. Therefore, objective
pathological diagnosis results can effectively help doctors choose the most appropriate
treatment mode, thereby improving the survival rate of patients. For the current problem of
incomplete experimental subjects in the computer-aided diagnosis of lung cancer subtypes,
this study included relatively rare lung adenosquamous carcinoma (ASC) samples for the first
time, and proposed a computer-aided diagnosis method based on histopathological images of
17
ASC, lung squamous cell carcinoma (LUSC) and small cell lung carcinoma (SCLC). Firstly,
the multidimensional features of 121 LC histopathological images were extracted, and then
the relevant features (Relief) algorithm was used for feature selection. The support vector
machines (SVMs) classifier was used to classify LC subtypes, and the receiver operating
characteristic (ROC) curve and area under the curve (AUC) were used to make it more intuitive
evaluate the generalization ability of the classifier. Finally, through a horizontal comparison
with a variety of mainstream classification models, experiments show that the classification
effect achieved by the Relief-SVM model is the best. The LUSC-ASC classification accuracy
was 73.91%, the LUSC-SCLC classification accuracy was 83.91% and the ASC-SCLC
classification accuracy was 73.67%. Our experimental results verify the potential of the
auxiliary diagnosis model constructed by machine learning (ML) in the diagnosis of LC.

2.3.1 ADVANTAGES AND DISADAVANTAGES

ADVANTAGES:
• Relief-SVM algorithm has the best classification performance in distinguishing lung
cancer subtypes, which provides a favourable guidance for the classification of lung
histopathological images
• Lung Cancer screening errors can be avoided
• Clinician's multifarious work pressure can be slowed

DISADVANTAGES:
• When faced with a small number of medical images, traditional manual methods have
shown great potential in the classification of lung histopathological images.
• The CNN models which are trained directly by using the histopathology image data set
and the fine-tuning pre-trained CNNs models and custom models , CNN model show
weak classification performance.

2.4 LDNNET: Towards Robust Classification of Lung Nodule and Cancer Using Lung
Dense Neural Network YING CHEN , YERONG WANG , FEI HU, LONGFENG
FENG,TAOHUI ZHOU, AND CHENG ZHENG

Lung nodule classification plays an important role in diagnosis of lung cancer which is
essential to patients’ survival. However, because the number of lung CT images in current
dataset is relatively small and the ratio of nodule samples to non-nodule samples is usually
very different, this makes the training of neural networks difficult and poor performance of
18
neural networks. Hence, LDNNET is proposed, which adopts Dense-Block, batch
normalization (BN) and dropout to cope with these problems. Meanwhile, LDNNET is an
adaptive architecture based on convnets combining Softmax classifier which is utilized to
alleviate the problems of training deep convnets. Follows are our main work: Firstly, we
utilized LDNNET on database Lung Nodule Analysis 2016 (LUNA16) for lung nodule
classification and database KAGGLE DATA-SCIENCE-BOWL-2017(Kaggle DSB 2017) for
lung cancer classification; Secondly, the comparison experiments are designed to compare the
performance of dense connection, pooling layer and the input pixel size of lung CT(Computed
Tomography) images; Thirdly, data enhancement, dense connection and dropout layer were
utilized in LDNNET to reduce overfitting; Fourthly, pre-processing methods, for instance
enhanced contrast, median filtering, Laplacian filtering are compared to the no-processing
method to explore the effect of pre-processing on lung CT images classification. Fifthly,
accuracy, specificity and sensitivity on LUNA16 are 0.988396, 0.994585 and 0.982072 and
these indicators on Kaggle DSB 2017 are 0.999480, 0.999652 and 0.998974. Furthermore,
AUC for both two datasets is over 0.98.

2.4.1 ADVANTAGES AND DISADAVANTAGES

ADVANTAGES:
• LDNNET network is robust for the reason that both in the classification of lung nodules
and the classification of lung cancer have both achieved good performance.

• The LDNNET network can achieve higher accuracy than other existing algorithms for
different pixel sizes of input lung CT images and different experimental targets.

• LDNNET does not use semantic segmentation and positioning, hence this network is
easier to be deployed and utilized under actual circumstances

DISADVANTAGES:
• LDNNET does not perform any pre-processing methods.

• It is effective in classifying only lung CT images and not brain CT images, liver CT
images and bone CT images.

19
2.5 Deep Learning Assisted Predict of Lung Cancer on Computed Tomography Images
Using the Adaptive Hierarchical Heuristic Mathematical Model HENG YU , ZHIQING
ZHOU , AND QIMING WANG (2020)

Lung cancer is known to be one of the most dangerous diseases which are the main reason for
disease and death when diagnosed in primitive stages. Since lung cancer can only be detected
more broadly after it spread to lung parts and the occurrence of lung cancer in the earlier stage
is very difficult to predict. It causes a greater risk as radiologists and specialist doctors assess
the existence of lung cancer. For this reason, it is important to build a smart and automatic
cancer prediction system that is accurate and at which stage of cancer or to improve the
accuracy of the previous cancer prediction that will help determines the type of treatment and
treatment depth depending on the severity of the disease. In this paper, the Adaptive
Hierarchical Heuristic Mathematical Model (AHHMM) has been proposed for the deep
learning approach. To analyse deep learning based on the historical therapy scheme in the
development of Non-Small Cell Lung Cancers (NSCLC) automated radiation adaptation
protocols that aim at optimizing local tumor regulation at lower rates of grade 2 RP2 radiation
pneumonitis. Furthermore, the system proposed consists of several steps including acquiring
the image, pre-processing, binarization, thresholding, and segmentation, extraction of features
and detection of deep neural network (DNN). Segmentation of the lung CT image is carried
out to extract any significant feature of a segmented image, and a specific feature extraction
method is implemented.
2.5.1 ADVANTAGES AND DISADVANTAGES
ADVANTAGES
• It uses the Modified K-means algorithm to pre-classify pictures into slices of images in
the same image where the DNN will concentrate on the image classification of images
in similar images
• The AHHMM system predicts CT scan images of lung cancer successfully.
• The findings of the evaluation showed that around 90% of the images has correctly
identified.

DISADVANTAGES
• The images has been divided into one convolutional layer only.
• It consists of several steps including acquiring the image, pre-processing, binarization,
thresholding, and segmentation, extraction of features and detection of deep neural
network (DNN) which increases the time complexity.
• It is difficult to classify the ’heavier’ images with Dense Neural Network (DNN).

20
 CHAPTER 3

SYSTEM DESIGN AND

IMPLEMENTATION

21
 3.1 SYSTEM REQUIREMENTS

3.1.1. Hardware requirements:

Processor : Intel-i3

Hard disk : minimum 120 GB

RAM : minimum 4 GB

3.1.2. Software Requirements:

Operating System : Windows / Linux

Simulation Tool : Anaconda with Jupyter Notebook

Libraries Required:
• TensorFlow: Just to use the tensor board to compare the loss and adam curve our result
data or obtained log.
• Keras: To pre-process the image dataset.

• Matplotlib: To display the result of our predictive outcome.

• OS: To access the file system to read the image from the train and test directory from our
machines.

TensorFlow:
TensorFlow is a Python library for fast numerical computing created and released by
Google. It is a foundation library that can be used to create Deep Learning models directly or
by using wrapper libraries that simplify the process built on top of TensorFlow.

22
 TensorFlow is a software library or framework, designed by the Google team to
implement machine learning and deep learning concepts in the easiest manner. It combines the
computational algebra of optimization techniques for easy calculation of many mathematical
expressions.

Let us now consider the following important features of TensorFlow −

• It includes a feature of that defines, optimizes and calculates mathematical expressions

easily with the help of multi-dimensional arrays called tensors.

• It includes a programming support of deep neural networks and machine learning

techniques.

• It includes a high scalable feature of computation with various data sets.

• TensorFlow uses GPU computing, automating management. It also includes a unique

feature of optimization of same memory and the data used.

Keras:

Keras runs on top of Open source machine libraries like TensorFlow, Theano or
Cognitive Toolkit (CNTK). Theano is a python library used for fast numerical computation
tasks. TensorFlow is the most famous symbolic math library used for creating neural networks
and deep learning models. TensorFlow is very flexible and the primary benefit is distributed
computing. CNTK is deep learning framework developed by Microsoft. It uses libraries such
as Python, C#, C++ or standalone machine learning toolkits. Theano and TensorFlow are very
powerful libraries but difficult to understand for creating neural networks.

Keras is based on minimal structure that provides a clean and easy way to create deep learning
models based on TensorFlow or Theano. Keras is designed to quickly define deep learning
models. Well, Keras is an optimal choice for deep learning applications.

Features

Keras leverages various optimization techniques to make high level neural network API easier
and more performant. It supports the following features −

23
 • Consistent, simple and extensible API.

• Minimal structure - easy to achieve the result without any frills.

• It supports multiple platforms and backends.

• It is user friendly framework which runs on both CPU and GPU.

• Highly scalability of computation.

Matplotlib:

Matplotlib is one of the most popular Python packages used for data visualization. It is
a cross-platform library for making 2D plots from data in arrays. Matplotlib is written in
Python and makes use of NumPy, the numerical mathematics extension of Python. It provides
an object-oriented API that helps in embedding plots in applications using Python GUI
toolkits such as PyQt, WxPythonotTkinter. It can be used in Python and IPython shells,
Jupyter notebook and web application servers also.

Matplotlib has a procedural interface named the Pylab, which is designed to resemble
MATLAB, a proprietary programming language developed by MathWorks. Matplotlib along
with NumPy can be considered as the Open source equivalent of MATLAB.

OS:

The OS module in Python comes with various functions that enables developers to
interact with the Operating system that they are currently working on. In this article we’ll be
learning mainly to create and delete a directory/folder, rename a directory and even basics of
file handling.

Python OS module provides the facility to establish the interaction between the user and
the operating system. It offers many useful OS functions that are used to perform OS-based
tasks and get related information about operating system.

The OS comes under Python's standard utility modules. This module offers a portable
way of using operating system dependent functionality.

24
3.2 EXISTING SYSTEM

Lung nodule malignancy prediction is an essential step in the early diagnosis

of lung cancer. Besides the difficulties commonly discussed, the challenges of this
task also come from the ambiguous labels provided by annotators, since deep learning
models have in some cases been found to reproduce or amplify human biases. In this
paper, we propose a multi-view ‘divide-and-rule’ (MV-DAR) model to learn from
both reliable and ambiguous annotations for lung nodule malignancy prediction on
chest CT scans. According to the consistency and reliability of their annotations, we
divide nodules into three sets: a consistent and reliable set (CR-Set), an inconsistent
set (IC-Set), and a low reliable set (LR-Set). We propose the MV-DAR model to learn
from ambiguous labels, including inconsistent malignancy scores given by multiple
radiologists and an unreliable single score given by only one radiologist, for lung
nodule malignancy prediction. Our results on the LIDC-IDRI dataset indicate that the
proposed MV-DAR model is substantially superior to four proxy-label based methods
and an identification-based method. Meanwhile, the ablation studies not only
demonstrate the effectiveness of each part of MV-DAR, but also confirm its
robustness of ambiguous labels.

3.2.1 DISADVANTAGES OF EXISTING SYSTEM:

• They are predicting Binary classification.

• Accuracy is very low.
• They are implementing the complex process.

3.3 PROPOSED SYSTEM

Lung cancer mainly affects older people. It’s rare in people younger than 40.
More than 4 out of 10 people diagnosed with lung cancer in the UK are aged 75 and
older. Although people who have never smoked can develop lung cancer, smoking
25
 is the most common cause. This is because smoking involves regularly inhaling a
number of different toxic substances. Adenocarcinoma is the most common type of
lung cancer in the United States and usually begins along the outer sections of the
lungs. It is also the most common type of lung cancer in people who have never
smoked. Squamous cell carcinoma is also called epidermoid carcinoma. It often
begins in the bronchi near the middle of the lungs. Proper CNN Algorithm should
be implemented in this project. Doctors observe the lung cancer types of patients,
and small changes on that reports are very tricky to understand. The proposed
system will predict lung cancer types effectively when compare to the doctors too.
We are going to classify three stages of cancer. Lung Benign Tissue, Lung
Adenocarcinoma, and Lung Squamous Cell Carcinoma. A different number of
images is collected from each class and trained by a proper CNN Architecture and
get accuracy from different architectures. First we are going to pre processing our
dataset, and implementing CNN architectures and comparing accuracy of all the
architectures and chosen the best architecture by the visual of more accuracy.
Finally deploying our model.

3.3.1 ADVANTAGE OF PROPOSED SYSTEM

• We predict the types of Lung Cancer.

• We are implementing CNN Algorithm for all the Architecture.
• Deployment is done here.

26
3.4 SYSTEM ARCHITECTURE:

27
 CHAPTER 4
MODULE DESCRIPTION

28
 4.1 OVERVIEW OF THE PROJECT

WORKING PROCESS OF LAYERS IN CNN MODEL:

A Convolutional Neural Network (ConvNet/CNN) is a Deep Learning algorithm

which can take in an input image, assign importance (learnable weights and biases) to various
aspects/objects in the image and be able to differentiate one from the other. The pre-processing
required in a ConvNet is much lower as compared to other classification algorithms. While in
primitive methods filters are hand-engineered, with enough training, ConvNets have the ability
to learn these filters/characteristics. The architecture of a ConvNet is analogous to that of the
connectivity pattern of Neurons in the Human Brain and was inspired by the organization of
the Visual Cortex. Individual neurons respond to stimuli only in a restricted region of the visual
field known as the Receptive Field. Their network consists of four layers with 1,024 input
units, 256 units in the first hidden layer, eight units in the second hidden layer, and two output
units.
Input Layer:

Input layer in CNN contain image data. Image data is represented by three dimensional
matrixes. It needs to reshape it into a single column. Suppose you have image of dimension 28
x 28 =784, it need to convert it into 784 x 1 before feeding into input.

Convo Layer:
Convo layer is sometimes called feature extractor layer because features of the image are
get extracted within this layer. First of all, a part of image is connected to Convo layer to
perform convolution operation as we saw earlier and calculating the dot product between
receptive field (it is a local region of the input image that has the same size as that of filter) and
the filter. Result of the operation is single integer of the output volume. Then the filter over the
next receptive field of the same input image by a Stride and do the same operation again. It will
repeat the same process again and again until it goes through the whole image. The output will
be the input for the next layer.

29
Pooling Layer:

Pooling layer is used to reduce the spatial volume of input image after convolution. It is
used between two convolution layers. If it applies FC after Convo layer without applying
pooling or max pooling, then it will be computationally expensive. So, the max pooling is only
way to reduce the spatial volume of input image. It has applied max pooling in single depth
slice with Stride of 2. It can observe the 4 x 4 dimension input is reducing to 2 x 2 dimensions.

Fully Connected Layer (FC):

Fully connected layer involves weights, biases, and neurons. It connects neurons in one
layer to neurons in another layer. It is used to classify images between different categories by
training.

Softmax / Logistic Layer:

Softmax or Logistic layer is the last layer of CNN. It resides at the end of FC layer.
Logistic is used for binary classification and Softmax is for multi-classification.

Output Layer:
Output layer contains the label which is in the form of one-hot encoded. Now you have
a good understanding of CNN.

4.2 MODULES

There are three modules involved in this proposal namely,

• Dataset Collection & Pre-processing

• Training and Testing the Data
• Classification Identification
• Deployment

30
4.2.1 DATASET COLLECTION & PRE-PROCESSING

We have to import our data set using Keras pre-processing image data generator
function also we create size, rescale, range, zoom range, horizontal flip. Then we import our
image dataset from folder through the data generator function. Here we set train, test, and
validation also we set target size, batch size and class-mode from this function we have to train
using our own created network by adding layers of CNN.

4.2.2 TRAINING (Feature Extraction) AND TESTING THE DATA

To train our dataset using classifier and fit generator function also we make training steps per
epoch’s then total number of epochs, validation data and validation steps using this data we
can train our dataset.

CNN Model Summary details

4.2.3 CLASSIFICATION IDENTIFICATION

We give input image using Keras pre-processing package. That input Image converted into
array value using pillow and image to array function package. We have already classified lung
histopathological in our dataset. It classifies what are the types of cancer. Then we have to
predict our lung histopathology using predict function.

31
 Given dataset

Lung
Feature CNN Model Histopathology
Input
Extractions Classification
image

The Lung Histopathology recognition method is based on a two-channel architecture

that is able to recognize Lung Histopathology types. The Lung Histopathology parts are
cropped and extracted and then used as the input into the inception layer of the CNN. The
Training phase involves the feature extraction and classification using convolution neural
network.

4.2.4 DEPLOYMENT
The User can Interact with the Web Content which is developed by Django framework.
The web Interface has the feature that the User can upload the histopathological image and
compare their image with Previous trained image. We will develop the website which has the
capability to evaluate or analyze the image whether the user has benign or malignant cancer.
4.3 UML DIAGRAMS
4.3.1 USE CASE DIAGRAM

32
In the above Use Case diagram people and end user are considered as Actors.
Collect the image data from people, after pre-processing using CNN architecture we can
analyze the Lung cancer.
4.3.2 CLASS DIAGRAM

Seven Classes have been used, Image has been collected and taken as input and data has been
tested using TensorFlow Model, further preprocessed an image and analyzed by Deep
learning.

33
4.3.3 SEQUENCE DIAGRAM

Image Collected from people and going under Pre-Processing, further checked with past data
(trained data) and lung pathology detected by DL method.

4.3.4 ER DIAGRAM

34
In the above Entity Relationship Diagram, Here Image, Frame work and CNN algorithm acts
as a Entity and TensorFlow Model , past data are Attributes. Pre-processing is the relationship
between Tuning Model and Deep Learning Algorithm and output has been finalized.

4.3.5 DATA FLOW DIAGRAM

Lung Histopathology
Dataset

Pre-processing Test
dataset

DFD LEVEL - 0

Lung Histopathology Dataset

Pre-processing Test
dataset

Training Dataset

DFD LEVEL -1

35
Lung Histopathology
Dataset

Pre-processing Test
dataset

CNN Algorithm Analysis of

Training Dataset Lung
Histopathology

DFD LEVEL -2

36
 CHAPTER 4
CONCLUSION AND FUTURE WORK

37
4.1 CONCLUSION

It focused how image from given dataset (trained dataset) and past data set used to
predict the pattern of Lung Histopathology using CNN model. This brings some of the
following insights about Lung Histopathology prediction. The major benefit of the CNN
classification framework is the ability to classify images automatically. The Lung
Histopathology mainly contribute to face misshape and often can’t be remedied because the
patients are diagnosed too late with the diseases. In this study, we have discussed the overview
of methodologies for detecting the abnormalities in Lung Histopathology images which
includes collection of Lung Histopathology image data set, pre-processing techniques, feature
extraction techniques and classification schemes.

4.2 FUTURE WORK:

➢ To automate this process by show the prediction result in web application or desktop
application.
➢ To optimize the work to implement Embedded System.

38