El-Bayan College

Biomedical Engineering

Pulmonary Nodule Detection

Using CT Image

Prepared By:
Mohamed Omer Hassan
Mojtaba Alhady Hassan
Abobakier Salah khairy
Mohamed Zain Elabdeen

Supervisor:
Dr. Nada Abdalla

October, 2019
 ‫اآليـــــة‬

‫قال تعالى‪:‬‬

‫ت ْالعَ ِلي ُم‬ ‫س ْب َحانَك ََل ِع ْل َملَنَا ِإ ََل َما َ‬

‫علَ ْمتَنَا ِإنَ َك أَ ْن َ‬ ‫﴿قَالُوا ُ‬
‫ْال َح ِكي ُم﴾‬

‫صدق هللا العظيم‬

 Acknowledgement
We extend our thanks to all who stood
With us to achieves this research which
It com because of grace of God
And reconcile.
We would like to give special thanks
Our supervisor
Dr. Nada Abdalla
For his great help and support.
Our teachers that gave us information and all staff in
Elbayan College.
Finally yet importantly we dedicate this project for
everyone that helped us to be at the place that we are today.
 Dedication
Dedication to my mother…
Whit warmth and faith...
Dedication to my father…
Whit love and respect …
Dedication to my friends..
Whom we cherish their friendship
Dedication to my special people
Who mean so much to me…
Dedication to all my teachers …
In whom I believe so much …
 Chapter One

Introduction

1.1 Preface

Lung cancer is the most common cause of death due to cancer in both men
and women throughout the world. Statistics from the American Cancer
Society estimated that in 2014 about 224,000 new cases of lung cancer in
the U.S. occurred and about 159,000 deaths were due to the disease.
According to the U.S. National Cancer Institute, approximately one out of
every 14 men and women in the U.S. will be diagnosed with cancer of the
lung at some point in their lifetime. Pulmonary nodules can be an
indication for primary lung cancer, CT offers better contrast, detect
smaller, earlier stage nodules with a higher sensitivity. CAD is
technologies to improve the quality and productivity of radiologists’ tasks
by improving the accuracy and consistency of radiological diagnoses and
also by reducing the image reading time and miss. The objective of this
study is identifying all nodules from the chest CT lung images and
classifying these nodules into Exist and dose not excite, to reduce the false
positive rate using Image processing and Neural Network techniques.

The usage of image processing in medical image solve the problem of

reduced accuracy of detection and diagnosis of dieses. Many dieses can be
detected by the image processing and this method improve the knowledge
and the give the ability to diagnose well. Medical images that used in
diagnosing and classification is free of compression in order to keep small
details inside the image since compression of images remove some image
characteristics, so BMP, DICM and TFT images are classified as medical
image format because it doesn’t have compression.

Many steps must be done in order to reach the last stage of classification,
starting from loading image, apply preprocessing, segmentation, feature
extraction and finally classification. Moreover many features can be used
to get accurate results such as

1.2 Problem Definition

Eye detection and diagnosis is the traditional method in detecting

Pulmonary Nodule and classification of region is usually done on CT scan
image, this method based on area and shape also

 Doctors vary in their knowledge.

 Eye diagnosis can lead to mistakes.
 Time consumed for diagnosis.
 Low Accuracy of detection and classification.

1.2 Proposed Solution

Prepare code using MTLAB code based on image processing technique

used to detected and classify pulmonary nodules in lung in order to reduce
the time and increase accuracy for classification and detection.

1.3 Project Objectives

The project aims and objectives represented into two categories:

1.3.1 General objective:

The objective from this project is using MTLAB code and image
processing technique to detected and classify pulmonary nodules in lung
in order to reduce the time and increase accuracy for classification and
detection.

 Provide a simple training program to increase the Doctors

knowledge on the pulmonary nodules detection and classification.
 Improve the Eye diagnosis and decrease the mistakes.
 Reduce time consumed for diagnosis.
 Provide high accuracy of detection and classification based on
computerized method.

1.3.2 Specific objective

 Study and analysis the Pulmonary Nodule.
 Study and analysis of Pulmonary Nodule preprocessing and
classification methods.

1.4 Thesis layouts

This project thesis consist of five chapter, in chapter one, some overview
including: Introduction, objective and problems statement was included,
while Chapter two represent the theoretical background and related works,
in Chapter three represent project methodology, while chapter four results
and discussion, in chapter five conclusion and recommendation was
included.
 Chapter Two

Literature Review

In this chapter a theoretical background was included along with the lung
anatomy and CT image with the way to apply detection of Pulmonary
Nodule.

2.1 Lung Anatomy

The lungs are the primary organs of the respiratory system in humans and
many other animals including a few fish and some snails. In mammals and
most other vertebrates, two lungs are located near the backbone on either
side of the heart. Their function in the respiratory system is to extract
oxygen from the atmosphere and transfer it into the bloodstream, and to
release carbon dioxide from the bloodstream into the atmosphere, in a
process of gas exchange. Respiration is driven by different muscular
systems in different species. Mammals, reptiles and birds use their
different muscles to support and foster breathing. In early tetrapods, air
was driven into the lungs by the pharyngeal muscles via buccal pumping,
a mechanism still seen in amphibians. In humans, the main muscle of
respiration that drives breathing is the diaphragm. The lungs also provide
airflow that makes vocal sounds including human speech possible.

Humans have two lungs, a right lung and a left lung. They are situated
within the thoracic cavity of the chest. The right lung is bigger than the
left, which shares space in the chest with the heart. The lungs together
weigh approximately 1.3 kilograms (2.9 lb), and the right is heavier. The
lungs are part of the lower respiratory tract that begins at the trachea and
branches into the bronchi and bronchioles, and which receive air breathed
in via the conducting zone. The conducting zone ends at the terminal
bronchioles. These divide into the respiratory bronchioles of the
respiratory zone which divide into alveolar ducts that give rise to the
microscopic alveoli, where gas exchange takes place. Together, the lungs
contain approximately 2,400 kilometers (1,500 mi) of airways and 300 to
500 million alveoli. Each lung is enclosed within a pleural sac which
allows the inner and outer walls to slide over each other whilst breathing
takes place, without much friction. This sac also divides each lung into
sections called lobes. The right lung has three lobes and the left has two.
The lobes are further divided into Broncho pulmonary segments and
lobules. The lungs have a unique blood supply, receiving deoxygenated
blood from the heart in the pulmonary circulation for the purposes of
receiving oxygen and releasing carbon dioxide, and a separate supply of
oxygenated blood to the tissue of the lungs, in the bronchial circulation.

The tissue of the lungs can be affected by a number of diseases, including

pneumonia and lung cancer. Chronic obstructive pulmonary disease
includes chronic bronchitis and previously termed emphysema, can be
related to smoking or exposure to harmful substances such as coal dust,
asbestos fibers and crystalline silica dust. Diseases such as bronchitis can
also affect the respiratory tract. In embryonic development, the lungs
begin to develop as an out pouching of the foregut, a tube which goes on
to form the upper part of the digestive system. When the lungs are formed
the fetus is held in the fluid-filled amniotic sac and so they do not function
to breathe. Blood is also diverted from the lungs through the ducts
arteriosus. At birth however, air begins to pass through the lungs, and the
diversionary duct closes, so that the lungs can begin to respire. The lungs
only fully develop in early childhood.

2.1.1 Structure of the human lungs

The lungs are located in the chest on either side of the heart in the rib cage.
They are conical in shape with a narrow rounded apex at the top, and a
broad concave base that rests on the convex surface of the diaphragm.[1]
The apex of the lung extends into the root of the neck, reaching shortly
above the level of the sternal end of the first rib. The lungs stretch from
close to the backbone in the rib cage to the front of the chest and
downwards from the lower part of the trachea to the diaphragm.[1] The left
lung shares space with the heart, and has an indentation in its border called
the cardiac notch of the left lung to accommodate this.[2][3] The front and
outer sides of the lungs face the ribs, which make light indentations on
their surfaces. The medial surfaces of the lungs face towards the centre of
the chest, and lie against the heart, great vessels, and the carina where the
trachea divides into the two main bronchi.[3] The cardiac impression is an
indentation formed on the surfaces of the lungs where they rest against the
heart.

Both lungs have a central recession called the hilum at the root of the lung,
where the blood vessels and airways pass into the lungs.[1] There are also
Broncho pulmonary lymph nodes on the hilum.[3]

The lungs are surrounded by the pulmonary pleurae. The pleurae are two
serous membranes; the outer parietal pleura lines the inner wall of the rib
cage and the inner visceral pleura directly lines the surface of the lungs.
Between the pleurae is a potential space called the pleural cavity
containing a thin layer of lubricating pleural fluid. Each lung is divided
into lobes by the in folding of the pleura as fissures. The fissures are double
folds of pleura that section the lungs and help in their expansion.[4]

Figure 2.1: Structure of the Human Lung

2.2 Breathing

Breathing is a combination of many factors. These factors are integrated

and controlled by the nervous system.(mechanics of breathing)The lungs
can be expanded and contracted by downward and upward movement of
the diaphragm to lengthen or shorten the chest cavity and by elevation and
depression of the ribs to increase and decrease the antero-posteriordia
meter of the chest cavity.
 Figure (2.2) mechanics of breathing

Surrounding each lung individually is a serious cavity, called the pleural

cavity. The minute quantity of serous fluid in the cavity serves as a
lubricant. This serves to minimize friction for the expansion and
contraction of the lungs during breathing. [6]
There are essentially two blood supplies for the lungs--nutrient blood and
functional blood. Nutrient blood is carried by the bronchial arteries from
the thoracic aorta. Nutrient blood provides nourishment and oxygen to
the tissues of the lung.
Functional blood is actually involved in the respiratory exchange of
gases between the alveoli and the capillaries. Functional blood is brought
to and from the lungs by the pulmonary cycle of the cardiovascular
system.[7]
2.3 Lung cancer [8]
Lung cancer is the uncontrolled growth of abnormal cells in one or both
of the lungs. While normal cells reproduce and develop into healthy lung
tissue, these abnormal cells reproduce faster and never grow into normal
lung tissue. Lumps of cancer cells (nodules) then form and grow. Besides
interfering with how the lung functions, cancer cells can spread from the
LUNG into the blood stream or Lymphatic system where they can spread
to other organs.
 Figure (2.3) several forms of lung cancer

Lung cancer is the most common cause of death due to cancer in both men
and women throughout the world. Statistics from the American Cancer
Society estimated that in 2014 about 224,000 new cases of lung cancer in
the U.S. occurred and about 159,000 deaths were due to the disease.
According to the U.S. National Cancer Institute, approximately one out of
every 14 men and women in the U.S. will be diagnosed with cancer of the
lung at some point in their lifetime.
2.4 Lung diseases [8]
The presence of certain diseases of the lung, notably chronic obstructive
pulmonary disease (COPD), is associated with an increased risk (four- to
six-fold the risk of a nonsmoker) for the development of lung cancer even
after the effects of concomitant cigarette smoking are excluded.
Pulmonary fibrosis (scarring of the lung) appears to increase the risk about
seven-fold, and this risk does not appear to be related to smoking.
2.5 Prior history of lung cancer [8]
Survivors of lung cancer have a greater risk of developing a second lung
cancer than the general population has of developing a first lung cancer.
Survivors of non-small cell lung cancers (NSCLCs, see below) have an
additive risk of 1%-2% per year for developing a second lung cancer. In
survivors of small cell lung cancers (SCLCs, see below), the risk for
development of second lung cancers approaches 6% per year.
2.6 Air pollution [8]
Air pollution from vehicles, industry, and power plants can raise the
likelihood of developing lung cancer in exposed individuals. Up to 1% of
lung cancer deaths are attributable to breathing polluted air, and experts
believe that prolonged exposure to highly polluted air can carry a risk for
the development of lung cancer similar to that of passive smoking.

2.7 Causes lung cancer [8]

2.7.1 Smoking: about 90% of lung cancers arising as a result of tobacco
use.
2.7.2 Passive Smoking: Research has shown that nonsmokers who reside
with a smoker have a 24% increase in risk for developing lung cancer
when compared with nonsmokers who do not reside with a smoker.
2.7.3 Asbestos fibers: are silicate fibers that can persist for a lifetime in
lung tissue following exposure to asbestos.
2.7.4 Radon gas: is a known cause of lung cancer, with an estimated 12%
of lung-cancer deaths in the U.S.
2.7.5 Familial predisposition : individual genetic susceptibility, may play
a role in the causation of lung cancer.
2.3 Pulmonary Nodule

Lung nodules can be cancerous, though most lung nodules are

noncancerous (benign). Lung nodules small masses of tissue in the lung
are quite common. They appear as round, white shadows on a chest X-ray
or computerized tomography (CT) scan. Lung nodules are usually about
0.2 inch (5 millimeters) to 1.2 inches (30 millimeters) in size. A larger lung
nodule, such as one that's 30 millimeters or larger, is more likely to be
cancerous than is a smaller lung nodule. If the doctor detects a lung nodule
on an imaging test, it's helpful to compare your current imaging scan with
a previous one. If the nodule on earlier images hasn't changed in size,
shape or appearance in two years, it's probably noncancerous.
Noncancerous lung nodules are often caused by previous infections.
Noncancerous lung nodules usually require no treatment. In some cases
doctors may recommend annual chest imaging to see if a lung nodule
grows or changes over time. If a lung nodule is new or has changed in size,
shape or appearance, your doctor may recommend further testing such as
a CT scan, positron emission tomography (PET) scan, and bronchus copy
or tissue biopsy to determine if it's cancerous.

Computed Tomography Scan

A CT scan or computed tomography scan (formerly computerized axial

tomography scan or CAT scan) makes use of computer-processed
combinations of many X-ray measurements taken from different angles to
produce cross-sectional (tomographic) images (virtual "slices") of specific
areas of a scanned object, allowing the user to see inside the object without
cutting.
Digital geometry processing is used to further generate a three-
dimensional volume of the inside of the object from a large series of two-
dimensional radiographic images taken around a single axis of rotation.[3]
Medical imaging is the most common application of X-ray CT. Its cross-
sectional images are used for diagnostic and therapeutic purposes in
various medical disciplines. The rest of this article discusses medical-
imaging X-ray CT; industrial applications of X-ray CT are discussed at
industrial computed tomography scanning.

The term "computed tomography" (CT) is often used to refer to X-ray CT,
because it is the most commonly known form. But, many other types of
CT exist, such as positron emission tomography (PET) and single-photon
emission computed tomography (SPECT). X-ray tomography, a
predecessor of CT, is one form of radiography, along with many other
forms of tomographic and non-tomographic radiography.

CT produces data that can be manipulated in order to demonstrate various

bodily structures based on their ability to absorb the X-ray beam.
Although, historically, the images generated were in the axial or transverse
plane, perpendicular to the long axis of the body, modern scanners allow
this volume of data to be reformatted in various planes or even as
volumetric (3D) representations of structures. Although most common in
medicine, CT is also used in other fields, such as nondestructive materials
testing. Another example is archaeological uses such as imaging the
contents of sarcophagi or ceramics. Individuals responsible for performing
CT exams are called radiographers or radiologic technologists.
Medical use

Since its introduction in the 1970s, CT has become an important tool in

medical imaging to supplement X-rays and medical ultrasonography. It
has more recently been used for preventive medicine or screening for
diseas

CT-Image for Lungs

CT scan can be used for detecting both acute and chronic changes in the
lung parenchyma, that is, the internals of the lungs. It is particularly
relevant here because normal two-dimensional X-rays do not show such
defects. A variety of techniques are used, depending on the suspected
abnormality. For evaluation of chronic interstitial processes (emphysema,
fibrosis, and so forth), thin sections with high spatial frequency
reconstructions are used; often scans are performed both in inspiration and
expiration. This special technique is called high resolution CT.

Advantages

There are several advantages that CT has over traditional 2D medical

radiography. First, CT completely eliminates the superimposition of
images of structures outside the area of interest. Second, because of the
inherent high-contrast resolution of CT, differences between tissues that
differ in physical density by less than 1% can be distinguished. Finally,
data from a single CT imaging procedure consisting of either multiple
contiguous or one helical scan can be viewed as images in the axial,
coronal, or sagittal planes, depending on the diagnostic task. This is
referred to as multiplanar reformatted imaging.
CT is regarded as a moderate- to high-radiation diagnostic technique. The
improved resolution of CT has permitted the development of new
investigations, which may have advantages; compared to conventional
radiography, for example, CT angiography avoids the invasive insertion
of a catheter. CT colonography (also known as virtual colonoscopy or VC
for short) is far more accurate than a barium enema for detection of
noduless, and uses a lower radiation dose. CT VC is increasingly being
used in the UK and US as a screening test for colon polyps and colon
cancer and can negate the need for a colonoscopy in some cases.

The radiation dose for a particular study depends on multiple factors:

volume scanned, patient build, number and type of scan sequences, and
desired resolution and image quality.[44] In addition, two helical CT
scanning parameters that can be adjusted easily and that have a profound
effect on radiation dose are tube current and pitch. Computed tomography
(CT) scan has been shown to be more accurate than radiographs in
evaluating anterior interbody fusion but may still over-read the extent of
fusion.

Noise in computed tomography

Often, noise is defined loosely as, the grainy appearance on cross-sectional

imaging; more often than not, this is quantum mottle. Noise in CT is
measured via the signal to noise ratio (SNR); comparing the level of
desired signal (photons) to the level of background noise (pixels deviating
from normal). The higher the ratio, the less noise is present on the image.
Noise on a cross-sectional image will equal a decrease in the picture
quality and inadvertently will hinder the contrast resolution. The best filter
is Gaussian filter because it’s faster.

Chapter Three

Methodology

In this chapter the research methodology is included along with the

algorithms used for preprocessing, segmentation, classification and
detection of Pulmonary Nodule in lung using SVM.

3.1 System block Diagram

The proposed work mainly gives a review that what steps are
performed throughout the entire process to detect Pulmonary Nodule
from CT-Image. The frame work is mainly consists of two phases. In
the first phase features extraction are extracted from CT-Image and in
the second phase Pulmonary Nodule is classified as Found or Not-
Found. All that done by MATLAB software. The steps of Pulmonary
Nodule detection shown in figure 4.1.
 Figure 3.1: The proposed model

In figure 4.1 the preprocessing stage is used for the filtering of noise from
image and including, adjust image contrast and etc.

a) Pre-Processing Stage: In the pre-processing stage three steps used:

CT Input Image: In this step, CT of the LUNG is given as the input. It has
the RGB (Red Green Blue) mixing present in it. It is the basic color used
in computers, television, graphics …etc.

Gray Scale Image: It is the second step of the pre-processing stage. Since
CT input image contains some RGB mixing in it, a clear expected output
cannot be obtained. Hence the input image is converted to gray scale image
which is the black and white image. This image ranges from 0 to 255. The
range 0 defines the black color and the range 255 defines the white color.
As the name indicates image may also contain the shades of grey.

Apply Filters: It is the third step of the preprocessing stage. The converted
gray scale image consist of some noise in the image .The output will not
be clear with the noise present in the gray scale image.

b) Image segmentation

Segmentation is the process of partitioning a digital image into multiple

segments (sets of pixels, also known as super pixels). The goal of
segmentation is to simplify and/or change the representation of an image
into something that is more meaningful and easier to analyze. Image
segmentation is typically used to locate objects and boundaries (lines,
curves, etc.) in images. More precisely, image segmentation is the process
of assigning a label to every pixel in an image such that pixels with the
same label share certain visual characteristics.

In case of medical image segmentation the aim is to:

 Study anatomical structure,

 Identify Region of Interest i.e. locate nodules, lesion and other
abnormalities,
 Measure tissue volume to measure growth of nodules (also decrease
in size of nodules with treatment).

Using segmentation in medical images is a very important task for

detecting the abnormalities, study and tracking progress of diseases and
surgery planning. Segmentation must not allow regions of the image to
overlap.

Thresholding based methods

Thresholding is one of the most generally used and oldest methods for
image segmentation. In thresholding approach, image segmentation is
based on gray level intensity value of pixels.

Histogram of image

Is consists of peaks and valleys, where each peak represents one region.
The valley between the peaks represents a threshold value. Histogram
thresholding method is based on a concept that divides the image into two
equal halves and histograms are compared to detect the nodules and
cropping method is used to find a proper physical dimension of nodules.
The threshold technique makes decision based on the local raw pixel
information. It helps in extracting the basic shape of an image, overlooking
the little unnecessary details. However, thresholding is often used as an
initial step in a sequence of image segmentation process. Its main
limitation is that only two classes are generated and it does not work when
confronted with structures that lack clear borders. Image segmentation
through thresholding is considered to be a simple and powerful approach
to segment the images that have light objects on dark background. On the
basis of thresholding value, there are two types of threshold values such
as global and local thresholding. The approach is called global
thresholding when the T is fixed or constant. Otherwise, it is called local
thresholding. If the background illumination is uneven, the global
thresholding is likely to fail. In local thresholding, multiple thresholds are
used to compensate the uneven illumination.

Spatial Filtering
The mechanics of spatial filtering are the process consists simply of
moving the filter mask from point to point in an image. The response of
the filter at that point is calculated using a predefined relationship. For
linear spatial filtering the response is given by a sum of products of the
filter coefficients and the corresponding image pixels in the area spanned
by the filter mask.
Include Smoothing filters : are used for blurring and for noise reduction.
Blurring is used in preprocessing steps, such as removal of small details
from an image prior to (large) object extraction, and bridging of small gaps
in lines or curves. Noise reduction can be accomplished by blurring with
a linear filter and also by nonlinear filtering. And Sharpening Spatial
Filters : The principal objective of sharpening is to highlight fine detail in
an image or to enhance detail that has been blurred, either in error or as a
natural effect of a particular method of image acquisition. Uses of image
sharpening vary and include applications ranging from electronic printing
and medical imaging to industrial inspection and autonomous guidance in
military systems.[10]

Edge Detection Based Methods

Edge-based methods are focused on detecting contour. They fail when the
image is blurry or too complex to identify a given border. The most
important feature in an image is the contrast. Contrast may be described
as discontinuities in the gray values of an image or variations in scene
illumination.

Gray Scale Image

The image was Gary scaled in order to convert the image from RGB to
block and white to limit the color of the object that required to be detected.

Feature extraction: In the feature extraction the image must first applied
to a segmentation algorithm, thresholding, edge detection was used to
detect the nodules area to be statically calculated using mean, median and
standard deviation.

Classification: here a comparison between the knowledge base statistical

value and the statistical values of the input image, this is done through
Support Vector Machine.
e) Diagnosis: the classification difference output is then compared to the
normal and abnormal samples.

Computer Model
Figure 3.2: System Flowchart
Mathematical Model

The final procedure of the proposed system is to confirm the suspicious

region and determine if it is a true nodule utilizing features obtained from
output.
Performance measures
Sensitivity (SE): The sensitivity of a test is the ability of the test to identify
correctly the affected individuals Proportion of persons testing positive
among affected individuals
𝑺𝑬=𝑻𝑷/F𝑷+𝑭𝑵 (3.1)
Specificity (SP): The specificity of a test is the ability of the test to identify
correctly non-affected individuals Proportion of persons testing negative
among non-affected individuals
𝑺𝑷=𝑻𝑵/𝑭𝑵+𝑻𝑵 (3.2)
Positive Predictive Value (PPV): Is the proportion of abnormal cases
Correctly identified
𝐏𝐏𝐕=𝐓𝐏/F𝐏+𝐅𝐏 (3.3)
Negative Predictive Value (NPV): Is the proportion of benign cases
correctly identified
𝐍𝐏𝐕=𝐓𝐍/𝐅𝐍+𝐓𝐍 (3.4)