Breast Cancer Classification Using Machine Learning. An Empirical Study


Diploma Thesis, 2020

77 Pages, Grade: 3.55


Excerpt

TABLE OF CONTENTS

DEDICATION

ACKNOWLEDGEMENTS

LIST OF TABLES

LIST OF FIGURES

ABSTRACT

CHAPTER ONE
INTRODUCTION
1.1 Background of the study
1.2 Statement of the Problem
1.3 Aim and Objective of the study
1.4 Research Questions
1.5 Significance of study
1.6 Scope of the study
1.7 Limitation of the study
1.8 Operational definition of terms

CHAPTER TWO
LITERATURE REVIEW
2.1 Overview of Breast Cancer
2.2 Risk Factors
2.3 Symptoms of breast cancer
2.4 Screening/Diagnosis of breast cancer
2.5 Treatment of breast cancer
2.6 Breast cancer type using system gaining knowledge
2.7 Diagnosis and Treatment of Breast Cancer
2.8 Advantages and Disadvantages of Machine learning
2.9 EMPIRICAL STUDIES

CHAPTER THREE
SYSTEM DESIGN AND ANALYSIS
3.1 Data Collection Methods
3.2 DESIGN LANGUAGES, TOOLS, AND TECHNIQUES OF THE PROPOSED SYSTEM
3.3 TECHNIQUES OF PROPOSED SYSTEM
3.4 ANALYSIS OF THE EXISTING SYSTEM
3.5 ANALYSIS OF THE PROPOSED SYSTEM
3.6 DESIGN OF THE PROPOSED SYSTEM
3.7 FUNCTIONALITY OF THE SYSTEM

CHAPTER FOUR
IMPLEMENTATION AND DOCUMENTATION
4.1 SYSTEM TESTING
4.2 Test Plan
4.3 Test Data
4.4 Test Result

CHAPTER FIVE
SUMMARY, CONCLUSION AND RECOMMENDATION
5.1 Summary
5.2 Conclusion
5.3 Recommendations
5.4 SUGGESTION FOR FUTURE STUDIES

REFRENCES

DEDICATION

This study is dedicated to God Almighty for his protection and love throughout the course of this program. To my late wife, Mrs. Ugwu Regina for her words of enragement and prayers. Rest on.

ACKNOWLEDGEMENTS

First and foremost, appreciation to God almighty for this infinity mercy, Grace, Protection and love throughout the course of the write up. May your name be praise forever.

I am grateful to my supervisor Mrs. Inyang, for her relentless efforts despites her busy schedule still went through and make necessary corrections in this project work. Thank you Ma.

I wish to thank the entire department of computer science, Benson Idahosa University, for their tolerance, impartation of knowledge and guidance throughout this postgraduate journey. God bless you all.

My unreserved appreciation goes to my lecturers Mr. Sam Obadan, Mrs. Grace Iyawe, Mrs. Inyang, Dr. Maxwell, Dr. K.O Obahiagbon, Mr. Igbinigie Prince-collins for the knowledge they have uniquely impacted on me throughout each stages of my studies in Benson Idahosa University.

To my late wife Mrs. Regina Peter, who has contributed immensely to my life within four years of our marriage, you have no idea of how frat I am, you led to Christ, you set me on fire and now you are not here to watch the burning. Your silent labour of love will forever be remembered.

To my loving kids, Fortune and Favor for being there for me all through, I love you guys. To my kid sister, Ugwu Patience, for taking care of my kids. God will grant your heart desire in Jesus name.

To my Mum, Mrs. Regina Ugwu, for her moral upbringing and words of encouragement and prayers, God bless you mum, you are the Best. To my Sister and Her Husband, Mr. & Mrs. Itodo Peter. To all Oche Family for their encouragement God bless you all.

I appreciate my friends and classmate, who in one way or the other contributed immensely to this achievement and finally, to everyone who is one way or the other contributed but not mention here, your support is duly acknowledged. God bless you all in Jesus name (Amen).

LIST OF TABLES

Table 3.1 Summary of the computer used

Table 4.2 Graphical view of the aforementioned metrics works

Table 3.3 Tree Method

Table 3.4 (X, Y and Z) with two possible axes

Table 4.1The summary of the report

LIST OF FIGURES

Fig 2.1: HER2-Positive Breast Cancer

Fig. 2.2: Hormone-Sensitive Breast Cancer

Fig. 2.3: Signs of Inflammatory Breast Cancer

Fig. 2.4 Classification of breast cancer and non-cancer cells from reduced feature set

Figure 3.1 Dataset with 569 entries, with 31 columns

Figure 3.2 Comparison of prog. Languages

Fig. 3.3: Data analysis 1

Fig. 3.4a Visualization of data

Figure 3.5: Tree Method

Figure 3.6 Separation between classes

Figure 3.7 Support Vector Machines (SVMs)

Figure 3.8 Functionality of machine learning

Fig. 4.1 Logistic regression Confusion Matrix

Fig. 4.2 Random Forest Confusion Matrix

Fig. 4.3 SVC Confusion Matrix

ABSTRACT

Breast cancer is the most typical of all varieties of disease and very dangerous to health specifically amongst women. The main aim of the study was to detect and access classification of breast cancer using machine learning who have good knowledge about the prevention of breast cancer and are ready to learn computerized way of solving the problem using machine learning system. Classification of breast cancer leads pathologists to find a systematic and objective prognostic, generally the most frequent classification is binary (benign cancer/malign cancer). Today, Machine Learning (ML) techniques are being broadly used in the breast cancer classification problem. They provide high classification accuracy and effective diagnostic capabilities. To have better understanding of the model developed, Wisconsin data set approach was adopted. Logistic regression was used in analyzing data set. Data set was arranged to a cutoff point at 0.5 which means that anything below 0.5 will be class 0 and anything above 0.5 will be class 1. Support vector machine was adopted using classification method. In testing, the dataset was split into training set which is 70% and the test set which is 30%. Although several ratios, such as 80:20, 75:25. The result was able to shows the classification report and the confusion matrix of all the algorithms used for the development of this classification.

CHAPTER ONE

INTRODUCTION

1.1 Background of the study

Breast cancer is the most typical of all varieties of disease and very dangerous to health specifically amongst women (Lakeshore, 2014). Report at the price of breast most cancers growth uncovers that one out of each eight women on this planet has a possibility of having the contamination in her existence time American Cancer Society (2015). In spite of the reality that the prevalence of breast most cancers boom is increasing anywhere at some point of the world, the pace of increment is higher in developing international locations in which past due popularity of infection is fundamental WHO (2014). Breast cancers is as of now a terrible medical issue in Nigeria with around 1 loss of life in each 25 found out cases (Olaleye, 2013). An enormous strain over breast cancer cases in Nigeria is the consistent ascent in the quantity of instances and passings, a circumstance which affirms Lakeshore Cancer Center expectation that bosom ailment cases may also ascend to forty two million by using 2020 in the both male and females inside the country (WHO, 2014).

The number one purpose for this growing mortality charge is due to loss of early detection of the ailment (Badar, 2013). This issue is always a right away consequence of loss of breast most cancers awareness discovered in maximum growing countries. Knowledge approximately breast cancer is a fundamental detail necessary for the early detection, prevention and treatment of this circumstance (Outlook, 2013). Knowledge for this is taken into consideration because the possession of correct knowledge of breast cancer, its signs and symptoms, threat factors, prevention, treatment options and facilities. Adequate expertise of breast most cancers will equip women with the potential to observe and discover signs before the sickness begins to unfold and are seeking for scientific assistance early; even as knowledge of causes and danger factors of most cancers will assist inside the prevention of this sickness by way of equipping to undertake preventive measures and suitable lifestyle modification.

Breast most cancers represents a chief health hassle at global degree representing the primary deadly cause for girls. Cancer prevention through consequent screening applications, early discovery and timely, stepped forward and diverse approach of treatment are commonly the most a hit approaches to lessen mortality. During the closing decade, the imaging systems registered a huge charge of progress, triggering a recurring screening that goes as speedy and as correct as feasible in detecting lesion characteristics. A huge range of technologies and contraptions developed based totally on X-rays evaluation, ultrasound evaluations and magnetic resonance techniques, among which mammography, echography and the magnetic resonance imaging offer the fine qualitative consequences in performance and input-output ratio (Caramihai, 2015).

Machine learning is a multidisciplinary field of take a look at that particularly concerned with the design of algorithms which permit computers to learn. The term “Machine Learning” comes from the synthetic intelligence network but now an afternoon it particularly the focusing place for many branches of engineering and technological know-how (Alpaydin 2014). Learning in particular refers to learning from facts or feature set. There are one-of-a-kind mastering strategies for statistical statistics evaluation. These are supervised getting to know, unsupervised getting to know and reinforcement mastering.

Machine learning turned into created to alternative human-like, natural organic, non-linear questioning inside the computerized world, sensible strategies are the most advanced modeling techniques that can compare and determine primarily based on an inference process that is just like human thinking and judging. The most largely implemented wise techniques (Dumitrache, 2015) are the synthetic smart techniques, the neural network techniques, the genetic algorithms.

1.2 Statement of the Problem

Cancer is one of the deadly illnesses that has threatened the sector. According to WHO (2014), approximately 12.5% of all deaths globally are because of cancer, with the percentage extra than the proportion of deaths as a result of HIV/AIDS, tuberculosis, and malaria put together (Anyaegbudike, 2012).

Therefore, the boom in the attacks and deaths of women with breast cancer in Edo State poses a pertinent question on the impact of breast cancer campaigns on Edo women with reference to their bad responses to early presentation of breast cancer. It is in view of the above and given the confirmation of American Cancer Society (2013) that breast most cancers deaths stay preventable on the early degree, that the researcher critically evaluated the effectiveness of the system mastering system to effortlessly classify which sort of breast most cancers (Benign or Malignant) is present.

The study will classify breast cancers into foremost problems: (Benign tumor and Malignant tumor). A benign tumor is a most cancers does now not invade its surrounding tissue or spread around the host. A malignant tumor is another kind of cancers which can invade its surrounding tissue or spread around the frame of the host. Benign cancers on uncommon event can also surely result in someone’s death, but as a fashionable rule they're no longer nearly as horrific because the malignant cancers. The malignant cancers at the contrary are like those killer bees. In this situation, you do not need to be doing something to them or maybe be everywhere near their hive, they will just spread out and attack you emass – they could even kill the individual if they are extreme enough (WHO, 2014).

Manual manner of cancer category into benign and malignant may be very tedious, susceptible to human error and unnecessarily time consuming. The proposed system while constructed can robotically classify the sort of most cancers into the safe (benign) and also the risky (malignant). This machine plays this role through the usage of machine getting to know algorithm. The following is the extensive of this new system: Classification mistakes could be notably removed, early analysis of disorder, removal of possible human mistakes and the device does no longer die. However, the researcher seeks to detect and assess the class of breast using Machine learning.

1.3 Aim and Objective of the study

The main objective is this study is to detect and access classification of breast cancer using machine learning

The specific objective of the study are:

1. To classify of breast cancer into benign and malignant using machine learning system

2. To detect breast cancer early using machine learning system

3. To eliminate breast cancer classification errors in the course of processing using machine learning system.

1.4 Research Questions

The research questions are as follows:

1. What is the classification of breast cancer into benign and malignant using machine learning system?
2. To what extend can you detect breast cancer early using machine learning system?
3. How do you eliminate breast cancer classification errors in the course of processing using machine learning system?

1.5 Significance of study

This study has both theoretical and practical significance. Theoretically, it will contribute to the articulation of the media campaign role in solving the problem of breast cancer using machine learning system. It will serve as a data base to computer science researchers who may be interested in learning the technological way of fighting breast cancer and future researchers, who may be embarking on similar research in future. Practically it will serve as a document for government and non-governmental organizations, policy makers and media campaigns planners in the field of breast cancer. To the medical field, this study will be of immense contribution to their daily clinical diagnostic approach as it will be more accurate and time saving to use machine learning to classify which type of breast cancer is present or not than relying on manual process which is time consuming and low accuracy.

1.6 Scope of the study

This study is restricted to all the health care worker in Igbinedion University Teaching Hospital, Okada who have good knowledge about the prevention of breast cancer and are ready to learn computerized way of solving the problem using machine learning system.

1.7 Limitation of the study

Problems are related to the over fitting and overgeneralization effects when using known facts and trying to generalize to other cases not described explicitly in the knowledge base. Such problems exist with methods that employ machine learning approaches too. Dearth of materials was also a great challenge to the present study.

1.8 Operational definition of terms

Machine learning is a multidisciplinary field of study that mainly concerned with the design of algorithms which allow computers to learn. The term “Machine Learning” comes from the artificial intelligence community but now a day it mainly the focusing area for many branches of engineering and science (Alpaydin et al., 2014).

Knowledge: is awareness and understanding that one has gained on nutrition during pregnancy through learning and practice and pregnant women was considered to be knowledgeable if she correctly answered greater than or equal to 70% of the total knowledge assessing questions.

Breast Cancer: Breast cancer is a malignant tumor which originates from breast tissue of women.

Prevention: The act of stopping breast cancer among women in Igbinedion university teaching hospital from happening using machine learning.

Diagnosis: The identification of the nature of an illness (breast cancer) or other problem by examination of the symptoms.

Treatment is the manner in which something or a disease is cared for or dealt with. An example of treatment is when someone is cared for very well. An example of treatment is when you are given antibiotics for your illness.

CHAPTER TWO

LITERATURE REVIEW

2.1 Overview of Breast Cancer

Breast most cancers is the most frequently diagnosed cancer in girls all over the world, accounting for nearly 1 % of all deaths and ranking the fifth most not unusual shape of most cancers and a prime cause of deaths amongst women of 30 years and above (Parkin et al., 2013).

Breast most cancers is a malignant tumor which originates from breast tissue. It is a cancer of the glandular breast tissue, in which the tissues are destroyed because of excessive boom of the most cancers cells, leading to the destruction of the encircling tissues and other organs thru the blood circulation (Russel et al., 2013). However, breast most cancers is ordinarily detected as a painless lump or mass of tissues called tumors (Blugs et al., 2015). The cancer cells typically start both in the cells of the lobules or the ducts. The lobules are the milk-generating glands at the same time as the ducts are the passages where the produced milk is collected from, to the nipples. The cancerous cells also can grow to be the stomal tissues consisting of the fatty and fibrous connective tissues of the breast, although this incidence isn't always commonplace. However in step with there are different degrees of breast most cancers that explicitly show how some distance the cancer cells have metastasized or unfold past the authentic tumor (Blugs et al., 2016).

Stages of Breast Cancer

Breast cancer levels are characterised by using the cancer length, the invasiveness or noninvasiveness of the most cancers, the lymph nodes and the metastasis of the cancer. Breast cancer ranges can also be defined as local, nearby and remote (Prsani et al., 2015).

Breast most cancers level can be nearby whilst the most cancers is restrained in the breast. It is local when the most cancers is within the lymph nodes, often within the armpit. While distant breast most cancers stage is in which the cancer has metastasized to different components of the frame (Prsani et al., 2015).

Moreover, TNM is another staging system used to explain most cancers. This incorporates the size of the tumor (T), the lymph node (N), and the unfold, or metastasis of the cancer to different components of the frame (M) (Prsani et al., 2015).

Stage 0

This degree explains non-invasive breast cancers as in DCIS (ductal carcinoma insitu). It is a level in which no evidence of cancer cells forming on any part of the breast invading neighboring normal tissue exists (WHO, 2014).

Stage I

Stage I portrays invasive breast most cancers invading normal surrounding breast tissue. In this stage, the tumor measures up to two cm and no lymph nodes are involved. There can also be a microscopic invasion in stage I breast cancer. In microscopic invasion, the most cancers cells just started out to invade the tissue outside the liner of the duct or lobule, but the invading most cancers cells do not measure extra than 1 mm (WHO, 2014).

Stage II

In this degree the breast cancer is growing, but it's miles nevertheless contained the breast or boom has simplest prolonged to the close by lymph nodes. This stage is split into companies: Stage 2A and Stage 2B. The difference is determined through the size of the tumor and whether the breast most cancers has spread to the lymph nodes (WHO, 2014).

Stage III

Stage III is divided into 3 categories referred to as IIIA, IIIB, and IIIC. Stage IIIA portrays the invasive breast most cancers with both no tumor but most cancers observed in auxiliary lymph nodes, clumped together to other systems, or may have unfold to lymph nodes close to the breastbone. Stage IIIB explains invasive breast cancer wherein the cancer involved is of any length and has unfold to the chest wall and/or skin of the breast and have additionally unfold to axillary lymph nodes, and sticking to different structures, close to the breastbone. For instance, inflammatory breast cancer is considered a degree IIIB instance with the standard features as: reddening of a big part of the breast skin, swollen or warmth feeling of the breast and spreading of the most cancers cells to the lymph nodes (WHO, 2014).

Stage IIIC explains invasive breast most cancers in which no symptom of breast cancer exists or wherein a lump of any size has spread to the chest wall and/or the skin of the breast (WHO, 2014).

Additionally, it includes the spreading of the cancer to lymph nodes above or below the collarbone as well as the cancer spread to the axillary lymph nodes close to the breastbone

Stage IV

In this level iv invasive breast most cancers has unfold past the breast and close by lymph nodes to different organs of the body, along with the lungs, distant lymph nodes, pores and skin, bones, liver, or brain. Advanced and “metastatic” are words used to describe level IV breast cancer (WHO, 2014). Cancer may be level IV in the beginning analysis or it may be a recurrence of a previous breast cancer that has metastasized to other elements of the frame. Breast cancer types with snap shots are:

1. HER2-Positive Breast Cancer

Abbildung in dieser Leseprobe nicht enthalten

Fig 1: HER2-Positive Breast Cancer

Source: (WHO, 2014).

In about 20% of patients, breast cancer cells have too many receptors for a protein called HER2. This type of cancer is known as HER2-positive, and it tends to spread faster than other forms of breast cancer. It's important to determine whether a tumor is HER2-positive, because there are special treatments for this form of cancer (WHO, 2014).

Abbildung in dieser Leseprobe nicht enthalten

Fig. 2: Hormone-Sensitive Breast Cancer

Source: (WHO, 2014).

Some types of breast cancer are fueled by the hormones estrogen or progesterone. A biopsy can reveal whether a tumor has receptors for estrogen (ER-positive) and/or progesterone (PR-positive). About two out of three breast cancers are hormone sensitive. There are several medications that keep the hormones from promoting further cancer growth (WHO, 2014).

Abbildung in dieser Leseprobe nicht enthalten

Fig. 3: Signs of Inflammatory Breast Cancer

Source: (WHO, 2014)

Inflammatory breast cancer is a rare, fast-growing type of cancer that often causes no distinct lump. Instead, breast skin may become thick, red, and may look pitted -- like an orange peel. The area may also feel warm or tender and have small bumps that look like a rash (WHO, 2014).

2.2 Risk Factors

Factors can fine be defined as something which can increase someone’s threat of developing breast most cancers. However, there are two corporations of hazard factors; the danger factors within ones manipulate and the kind, past ones manipulate. Some of the danger elements one can't control have been recognized as age, family records, and medical history whilst weight, physical activity and alcohols consumption are examples of risk elements you may control, (Boffetta et al., 2011)

2.2.1 Risk Factors within Control

i. Weight- overweight can make a contribution to at least one growing breast cancer due to excess fats tissue that breeds better estrogen, accountable for growth in ones danger of contracting breast most cancers. The extra fat is due to menopausal stage of the woman.
ii. Diet- Researchers believed that some weight loss plan can growth the possibility of having breast cancer. Diets like beef, animal fat that include diary fat in cheese, milk, and ice cream; which would possibly incorporate sure hormones that are negative to health. While a few believe that an excessive amount of cholesterol is a chance aspect however recommended a weight-reduction plan rich in vegetables and end result (ACS, 2013)
iii. Exercise- Lack of exercising can growth the threat of breast cancer even as engaging in 45-60 minutes exercise for about 5 or greater days every week in keeping with American Cancer Society can reduce the danger of breast cancer.
iv. Alcohol and smoking – It has been observed out that partakers of alcohol and smoking are at risk of agreement breast most cancers greater than non-partakers. The unfavourable impact of alcohol can limit ones liver’s capacity to manipulate blood ranges of the estrogen hormone (WHO, 2014)

2.2.2 Risk Factors beyond Control

1. Age – Age has been attributed the second biggest threat factors of breast most cancers.

i. Studies have proven that women from age 30-39 have 1 in 233 or 43% probabilities of developing breast cancer whilst at 60 years of age, danger increases to one in 27 or four% possibilities of contracting the sickness. In different words about ninety five% of the illnesses are expected on the genetic abnormalities that take location as an aftermath of ageing technique in addition to “put on and tear of existence (WHO, 2014).
ii. Gender/Sex – This is ready being a female. Researchers have confirmed gender because the primary danger component of breast most cancers. Though it's been located in men, however is one hundred% more in girls than in guys, particularly due to the activities of estrogen and progesterone, (female hormones) that make the lady breast revel in non-stop changes and growths.
iii. Race – It has also been find out that white girls are pretty susceptible to getting the sickness than Asian, Hispanic, Native American, African American women.
iv. Family history of breast cancer – This moreover has been attributed to causing women with own family history of mothers, daughters and sisters with breast most cancers to stand a better danger of getting the disorder (WHO, 2014)
v. Pregnancy and Breast Feeding. However, Pregnancy and breast feeding are extensively believed to lessen destiny assaults of breast most cancers because of carrying out a larger duration of breast feeding, together with from 1 to two years. While women who were given their first being pregnant on the age of 30, are at a better risk of breast cancer (Ogbodo et al., 2015)

2.3 Symptoms of breast cancer

Symptoms of breast cancer within the view of American Cancer Society, (2013) and Medical Women’s Association of Nigeria, (2011), best display bodily signs and symptoms of a painless lump or tumor on the early treatable level whilst the cancerous cells in question exhibit no signs due to the smallness of the lump. It then implies that the symptoms of the disorder at the later level of the cancerous cells include:

i. Lump or thickening within the breast, whether or not soft or not.
ii. Changes in size and shapes of the breasts.
iii. Depressions on the floor of the breasts
iv. Rashes or scaling of the skin
v. Drawing in of the nipples
vi. Newly seen veins
vii. Nipple Discharge or bleeding from the nipple

2.4 Screening/Diagnosis of breast cancer

Screening and analysis are very critical inside the early detection and remedy of breast cancer (Gallagher et al., 2011).

Breast most cancers screening consequently refers to testing an in any other case healthy woman for breast most cancers in an try to gain an early analysis, which has been set up to substantially improve outcomes or the chances of a hit remedy and survival (Blugs et al., 2011).

It is a system whereby girls study their breast via themselves with a view to come across any ordinary lump or swelling for set off scientific assistance. This is known as breast exam. (Kayode et al., 2015). While CBE is the clinically examination of breast for the prognosis of cancer cells. In the same vein, mammography screening is using X-rays to examine the breast for any uncharacteristic hundreds or lumps. Moreover; couple with the aforementioned screening gear, Giordano (2014) affirms the use and inclusion of Fine Needle Aspiration and Cytology (FNAC), for breast cancer diagnosis.

Nevertheless, mammography has been endorsed for women adherence because the simplest method of early detection and reduction in breast most cancers mortality (Gallagher et al., 2012)

2.5 Treatment of breast cancer

Breast cancer treatment depends at the sort of cancer, the level of most cancers, age, fitness repute and extra private characteristics (Blugs et al., 2016). The remedy may be done by surgical operation, radiotherapy (radiation), chemotherapy and tablets. There isn't any single treatment of breast most cancers however a mixture of the aforementioned treatment plans. Often, surgical procedure is employed at the early stage of the cancer even as chemotherapy is implemented at the superior level of the cancer.

2.6 Breast cancer type using system gaining knowledge

During the past few years, numerous contributions were made in literature concerning the software of sample reputation strategies for breast most cancers diagnosis in tissue level. Rejani and his organization proposed a sample popularity method to categorise the breast tumor (Rejani et al., 2017). They used the photo segmentation to segment the breast tissue corresponding to the tumor and used the discrete wavelet rework (DWT) as a function extraction approach to extract numerous capabilities from the segmented images. Then they also used SVM classifier to classify the breast tissue similar to the capabilities and completed an accuracy of 88.Seventy five%. Martin and his organization proposed the approach for detection of mass on digitized mammograms (Martins et al., 2014).

They used K-approach clustering algorithm for photograph segmentation and grey degree co-incidence matrix to explain and examine the texture of segmented systems in the photo. The class of these systems was finished through Support Vector Machines, which separate them into two groups; using shape and texture descriptors: hundreds and non-masses. The classification accuracy acquired from that method was eighty five%.

Karabatak and his organization proposed an automatic diagnosis-based pattern reputation system for detecting breast cancer primarily based at the association policies (AR) and Artificial neural community (ANN) (Karabatak and Ince 2014). In that examine, they used AR technique for reducing the measurement of breast most cancers database and ANN for clever category. The proposed machine i.E. The combination of AR and ANN, overall performance was as compared with most effective ANN version. The measurement of enter characteristic area became reduced from 9 to 4 by means of using AR. In the checking out level, they use three-fold move validation method to the WBCD to assess the proposed pattern popularity device performances. The correct type rate acquired from that AR + ANN system became ninety five.6%. Jele and his group proposed a framework for computerized malignancy grading of the first-class needle aspiration biopsy tissue (Jele et al., 2018). They used an SVM classifier to assign a malignancy grade based on pre extracted features, with accuracy up to 94.24%. Arodz and his organization proposed the Pattern reputation machine for computerized detection of suspicious looking anomalies in mammograms (Arodź et al., 2015). They used adaboost algorithm based classifier and completed an accuracy of 90%.Brook and his group proposed a method for Breast Cancer Diagnosis from Biopsy Images using SVM (Brook et al., 2015). They implemented multi-class SVM on standard function vectors to attain a excessive recognition price.

Fatima and his group used an approach for classifying the breast most cancers from Wisconsin breast most cancers prognosis (WBCD) database the usage of adaptive neuro-fuzzy inference device (ANFIS) (Fatima and Amine 2012). By the use of the ANFIS classifier they had been performed an accuracy of ninety eight.25 % in tissue stage. Mousa and his group proposed a pattern reputation method, to categorise masses for micro calcification and odd severity along with benign or malignant from mammographic image (Mousa et al., 2015). They used wavelet analysis as a characteristic extraction approach and fuzzy-Neuro as a classifier to obtain a better type price. Niwas and his organization proposed the pattern popularity undertaking; by thinking about Color Wavelet Features as feature extraction technique from segmented histopathology photo (Issac Niwas et al., 2012). They used SVM classifier which gives an accuracy of ninety eight.3%.

Akay proposed some other pattern reputation method by means of thinking about SVM classifier to classify benign and malignant loads (Akay 2014). They used WBCD breast most cancers database and carried out an accuracy of 98%. Shi and his group proposed some other approach for detection and classification of hundreds from breast ultrasound snap shots. They used textural capabilities, fractal features as characteristic extraction strategies and SVM, fuzzy assist vector system (FSVM) as the classifiers, to categorise the benign and malignant loads (Shi et al., 2015).After type, they have been carried out a type accuracy of ninety six.Four% for FSVM classifier. Schaefer and his institution proposed a pattern popularity approach, by using thinking about primary statistical functions, histogram capabilities, pass cooccurrence matrix features and mutual information based capabilities because the feature extraction approach and Fuzzy rule based classifiers as class approach to categorise the benign and malignant tumor types from the thermogram pix (Schaefer et al., 2014). They performed a classification fee of seventy nine.53% with the help of 14 partitions based fuzzy rule classifier.

Abbildung in dieser Leseprobe nicht enthalten

Fig. 2.4 Classification of breast cancer and non-cancer cells from reduced feature set

(Source: Almadi, 2013).

In the choice-making method for CN diagnosis, chance elements are taken into consideration essential when an atypical glandular cell (AGC) happens, since a analysis can inspire a debate among professionals, and this may sooner or later affect the clinical technique accompanied by means of docs that would lead to false positives or false negatives. As a result, the expert device begins with a fuzzy logics version that determines if a affected person is susceptible to growing CN consistent with their scientific information.

The fuzzy rules integrate one or extra fuzzy enter units (history) and establishe affiliation with output fuzzy sets (outcome chance). This rule layout is known as the Mamdani (Mamdani FLC (Fuzzy Logic Controler) proposed by means of Mamdani and Assilian in 1994. This controller uses the error e(okay) and the trade of error Δe(okay) to supply changes in the output feature of the controller.) kind and works with a fuzzy controller that settles up a system on its paintings region.

Cancer Manifestation

Cancer is a pathologic tissue increase originated because of a continual proliferation of peculiar cells. It can stem from any sort of cellular from one-of-a-kind tissues in the human organism. There are many styles of most cancers. One of the maximum threatening to women is cervical cancer (CN), which not like the others, if it's far detected throughout its early levels, the possibilities for a total healing are masses. More than 80% of deaths due to CN, come from nations with high or medium tiers of poverty. It is to be expected that the once a year figure of deaths due to CN will raise and exceed to eleven million by means of 2030 (Ahmadi, 2013).

Cervical most cancers's underlying purpose is the contamination by using the human papillomavirus (CANCER), which is a common sexually transmitted sickness (STD). However, 10 or twenty years are wished for a precursor harm produced via CANCER to be became invasive most cancers. Unfortunately, it's far anticipated that 95% of women (within the reproductive age) who inhabit growing countries have by no means taken a Pap smear (Eldeib, 2016).

The overall healing rate for this disorder is closely related to the degree of improvement at the moment of the diagnosis and to the supply of its treatment, since CN is lethal if it stays untreated. Due to its complexity, CN remedy immediately relies on its appropriate and correct prognosis.

2.7 Diagnosis and Treatment of Breast Cancer

A carcinogen breast tumor is a breast mass that is growing abnormally and out of control. There are three famous techniques for breast cancer diagnosis: mammography; FNA with visible interpretation; and surgical biopsy. The capability of these methods to diagnose cancer effectively when the disease is gift is: mammogram - from sixty eight% to seventy nine%; FNA with visible-interpretation - from sixty five% to ninety eight%; and surgical biopsy - one hundred%. It is cited that: mammography lacks sensitivity; the sensitivity of FNA with visible interpretation varies greatly (as a result of the visual interpretation); and even though surgical biopsy is correct it is also a very intrusive, time-ingesting and high-priced technique (Zemouri et al., 2018).

FNA, which has been extensively prevalent within the approach to investigating mammary lesions, is the easiest and fastest biopsy method to be finished, being a percutaneous process (through the skin) wherein the specialist doctor uses a thin needle (which varies from 0.6 to 0.8 mm) and a syringe to take samples of fluid from a breast cyst or take away clusters of cells in a strong mass. The needle is inserted into the pores and skin in the direction of the lesion, with the objective of gathering cells for similarly assessment in their morphology, quantity and distribution via cytological examination (Zemouri et al., 2018).

The genetic fabric extracted from the breast via FNA is typically despatched to a Pathology laboratory for exam by way of pathologists (medical doctors specialised in disorder analysis through lab testing), who carry out the evaluation figuring out the cells’ characteristics from watching, below a microscope, smears made with this cloth on sheets of glass and stained using special techniques (Watson et al., 2018).

There are instances that despite the fact that cytology is tremendous, a patient does not show malignant or precursor injuries. In different cases, cytology is bad, and in further checking, malignant anatomical adjustments are determined which aren't identified before everything, incurring in fake positives or fake negatives (Zemouri et al., 2018).

The current advised terminology for reporting outcomes in cervical cytology is the Bethesda system. This device was evolved for the National Cancer Institute (NCI) in 1988, with the reason of supplying a uniformed terminology to facilitate verbal exchange among the pathologist and gynecologist. The principal motive of this system is to inform the gynecologist the most facts to be had for use in a patients' remedy by the means of a descriptive report in which all of the cytological elements ought to be covered (hormonal, morphologic, and microbiological stages) (Karthik et al., 2014).

Treatment

Breast most cancers is typically treated with surgical operation, which may be observed by way of chemotherapy or radiation remedy, or both. A multidisciplinary method is most desirable. Hormone receptor-superb cancers are frequently dealt with hormone-blockading therapy over guides of several years. Monoclonal antibodies, or other immune-modulating remedies, may be administered in certain instances of metastatic and different superior levels of breast most cancers. Although this range of treatment remains being studied (Saini et al., 2015).

Surgery

Chest after proper breast mastectomy Surgery includes the physical elimination of the tumor, generally together with a number of the encompassing tissue. One or more lymph nodes can be biopsied at some stage in the surgical procedure; more and more the lymph node sampling is completed with the aid of a sentinel lymph node biopsy.

Standard surgical procedures encompass:

1. Mastectomy: Removal of the complete breast.
2. Quadrantectomy: Removal of one-region of the breast.
3. Lumpectomy: Removal of a small a part of the breast.

Once the tumor has been eliminated, if the man or woman desires, breast reconstruction surgical operation, a kind of plastic surgical treatment, may then be performed to enhance the classy appearance of the treated web site. Alternatively, girls use breast prostheses to simulate a breast below apparel, or pick a flat chest. Nipple prosthesis may be used at any time following the mastectomy (Leite et al., 2018).

Medication

Medications used after and further to surgical treatment are called adjuvant therapy. Chemotherapy or different sorts of remedy previous to surgical procedure are referred to as neoadjuvant therapy. Aspirin may additionally lessen mortality from breast cancer while used with other treatments (Leite et al., 2018).

There are presently three fundamental companies of medicinal drugs used for adjuvant breast most cancers remedy: hormone-blocking off dealers, chemotherapy, and monoclonal antibodies.

Hormonal remedy

Some breast cancers require estrogen to hold developing. They may be recognized by means of the presence of estrogen receptors (ER+) and progesterone receptors (PR+) on their floor (occasionally referred to together as hormone receptors). These ER+ cancers may be handled with capsules that both block the receptors, e.G. Tamoxifen, or as a substitute block the production of estrogen with an aromatase inhibitor, e.G. Anastrozole or letrozole. The use of tamoxifen is suggested for 10 years. Letrozole is recommended for five years. Aromatase inhibitors are handiest appropriate for women after menopause; but, in this institution, they seem better than tamoxifen. (Petit, 2014) This is because the lively aromatase in postmenopausal women isn't like the everyday form in premenopausal women, and consequently those dealers are ineffective in inhibiting the most important aromatase of premenopausal women. Aromatase inhibitors must now not accept to premenopausal girls with intact ovarian function (except they're additionally on remedy to forestall their ovaries from running). CDK inhibitors can be utilized in aggregate with endocrine or aromatase remedy (Leite et al., 2018).

Chemotherapy

Chemotherapy is predominantly used for instances of breast cancer in stages 2–4, and is especially useful in estrogen receptor-negative (ER-) disorder. The chemotherapy medicines are administered in combos, usually for durations of three–6 months. One of the maximum common regimens, referred to as "AC", combines cyclophosphamide with doxorubicin. Sometimes a taxane drug, including docetaxel, is delivered, and the regime is then referred to as "CAT". Another commonplace treatment is cyclophosphamide, methotrexate, and fluorouracil (or "CMF"). Most chemotherapy medicines work via destroying fast-developing and/or speedy-replicating cancer cells, either by inflicting DNA damage upon replication or by different mechanisms. However, the medications additionally damage speedy-growing ordinary cells, which can also cause critical facet results. Damage to the coronary heart muscle is the most risky difficulty of doxorubicin, for example (Leite et al., 2018).

[...]

Excerpt out of 77 pages

Details

Title
Breast Cancer Classification Using Machine Learning. An Empirical Study
Course
Computer Science
Grade
3.55
Author
Year
2020
Pages
77
Catalog Number
V1012996
ISBN (eBook)
9783346404824
ISBN (Book)
9783346404831
Language
English
Tags
breast, cancer, classification, using, machine, learning, empirical, study
Quote paper
Akor Ugwu (Author), 2020, Breast Cancer Classification Using Machine Learning. An Empirical Study, Munich, GRIN Verlag, https://www.grin.com/document/1012996

Comments

  • No comments yet.
Read the ebook
Title: Breast Cancer Classification Using Machine Learning. An Empirical Study



Upload papers

Your term paper / thesis:

- Publication as eBook and book
- High royalties for the sales
- Completely free - with ISBN
- It only takes five minutes
- Every paper finds readers

Publish now - it's free