CHOOSE YOUR CURRENCY


BREAST CANCER CLASSIFICATION USING MODIFIED CONVOLUTION NEUTRAL NETWORK TECHNIQUE

Amount: ₦8,000.00 |

Format: Ms Word |

1-5 chapters |



ABSTRACT

Cancer is the abnormal growth of cells in the body tissue which can spread out rapidly in any part of the body. Some common types include lung cancer, colorectum cancer, prostate cancer and breast cancer, among which breast cancer is the most predominant. Breast cancer forms and grow uncontrollably in the surrounding tissue of the breast. The disease has resulted in several deaths most due to lack of early diagnosis and treatment. In breast cancer diagnosis, accurate detection and classification of the disease have been a major challenge over the years. A lot of time is wasted in making diagnoses due to huge volume of images to analyze as a result of numbers of increasing cancer cases. Some tissues have similar characteristics and formation often occurs in clusters varying between 0.05mm – 1mm in size making it difficult to locate, as a result, classification accuracy rate tends to decrease.  Giant  strides  have  been  made  in  improving  detection  accuracy  through application of machine learning algorithms like support vector machines, decision trees which have slightly improved system accuracy but limited when operating on raw image data. It requires features from the image to be first extracted before being fed into the model. Therefore, there is a need to design a system that can automatically learn features and make accurate prediction. The research focus on the use of a more recent approach in medical image analysis using a modified Convolutional Neutral Network (CNN) technique to learn, detect and classify the presence of breast cancer whether they are malignant (abnormal tissue) or benign (normal tissue) with a high precision. Transferring learning approach is adapted instead of building model from the scratch a method which has been proven to work satisfactory in breast cancer classification tasks. The techniques allow one to  custom  model tailored  to  a  particular task  using popular architecture as a  baseline structure. The model uses the AlexNet architecture which was modified and tailored to our task. The work uses reflection and rotation as a form of augmentation technique to increase the amount of dataset that is used to train the model. The dataset undergoes a processing operation in order to enhance the low quality of the images before being fed into the model as training sources. The performance of the proposed model on test dataset is found to be; 95.80%,  95.00%,  80.00%,  92.30%  and  93.63%  for  accuracy,  sensitivity,  specificity, precision and F1score respectively. The results show significant improvement in classification accuracy when compared with existing literature using deep learning techniques and MIAS breast cancer dataset. This work will help doctors in making accurate classification and reduced time wastage that is usually associated with the manual ways of analyzing breast cancer.

CHAPTER ONE

1.0      INTRODUCTION

1.1      Background to the Study

Cancer causes an abnormal growth of cells and often spread out rapidly in other part of the body tissues. It is usually named after the part of the body in which it occurs. At the early stage of development, it forms a lump or mass, often referred to as tumor (Saeed et al., 2021). The lump produces no significant pain when it is treated early. Hence, conducting screening is crucial for early detection of the disease formation (Saira et al., 2018). Cancer occurs in both men and women but women are more likely susceptible to it, especially the breast cancer, which is considered as the most common type of cancer all over the world (Simon et al., 2020). According to the estimate made by American Cancer Society (ACS) about 1,898,160 new cases of invasive cancer were diagnosed for both women and men and 608,570 cancer deaths was recorded  as reported on U.S  Breast  Cancer Statistics, (Rebecca et al., 2021). Each year, approximately 14.1 million people suffer from breast cancer worldwide which results to 8.2 million deaths. It is expected that by the year 2025, there will be about 19.3 million newly reported cancer cases globally.

Most of the newly reported cancer cases occurred in developing nations, (Bray et al., 2018) which is expected to be on an increasing rate due to lack of awareness of the risk factors such  as  population  growth,  age  and  other  factors  such  environmental  toxins,  medical history, family history, genetics, exposure to radiation and infection. The accurate cancer statistics of Nigeria is relatively unknown due to lack of adequate funding by cancer registries, although some research studies have been conducted in different parts of the country. Jedy-Agba et al., (2018) reported Age Standardized incidence Rate (ASR) for all invasive cancer as 58.3/100,000 men and138.6/100,000 women while 66.4/100,000 men, 130.6/100,000 women are noninvasive cancer rate as conducted in Abuja and Ibadan based cancer registries, respectively.

In 2018, the World Health Organization (WHO) conducted a comprehensive study on cancer statistics of Nigeria, including both sexes and age, stated that there are about 115,950 new cancer cases and 70, 327 recorded death cases. The study also shows a glance of other types of cancer cases in the country represented in a form of pie chart in the Figure1.1.

Figure 1.1: International Agency for Research on Cancer GLOBOCAN (Bray et al., 2018) One of the ways of preventing an abnormal growth of cells (benign tumor) from becoming a full-blown cancer (malignant tumor) is by early detection, (Burcu et al., 2018). A lot needs to be done to educate women about the importance of self-examination and the significance of early disease detection. Over the years, doctors have made giant strides in early diagnosis and treatment of the disease and in reducing breast cancer mortality rate by using medical diagnostic imaging tools such as computer-aided detection, Digital mammography, Magnetic resonance imaging and ultrasound. Among which digital mammography procedures is considered to be faster and more accurate in detection, (Susan et al., 2017). A mammogram is an x-ray of breast indicating changes in breast tissues that can result to cancer. It helps in detecting very small lumps that can rarely be seen by the human eyes. In the past years, researchers have tried to automate the procedure by reducing the huge task of the radiologist in breast cancer diagnosis. No matter the expertise of the radiologists in examining mammograms, other limiting factors such as eye fatigue, interpretational errors, distractions need to be minimized as such could lead to false interpretations making the radiologist to give description that can harm the patients. For instance,  breast  biopsies  are  usually  prescribed  to  patient  diagnosed  as  cancerous. However,  over  40  –  60%  of  extraction  of  cells  in  breast  tissue  is  diagnosed  as noncancerous, which betray the necessity for correct mammography diagnosis (Ragab et al., 2019). On the other hand, cancer to be reported while in reality there is no malignant tissue this could lead to unnecessary treatment admitted to the patient all together as reported by (Elter and Horsch 2019).

Recent studies show that the use of Computer-Assisted Detection (CAD) software can greatly reduce the number of false interpretations and thereby increasing the accuracy of digital mammography diagnosis by Shen et al., (2020). Accuracy improvement in detection and classification of breast cancer is still a major challenge as reported in the work of (Saira et al., 2018). The focus behind this research is to design a model that can accurately detect  and  classify  breast  cancer  with  higher  degree  of  accuracy  using  modified Convolution Neural Network Technique. This can help the radiologist to make accurate diagnoses. This can ultimately, prevent late treatment due to false negatives and as well unnecessary treatments in cases of false positives.

1.2      Statement of the Research Problem

Breast cancer is common among women, leading to the major causes of death. Most of the death cases result from lack of early detection and diagnosis. Over the years, researchers have made great effort in early diagnosis and treatment of the disease using medical diagnostic tools like digital mammography. Mammogram images are formed by varying dissimilar factors like background interference, lightning variation and illumination changes (Suresh et al., 2018). The tumor formation often occurs in clusters varying between 0.05mm – 1mm in size which may also be located in a dense tissue, this posed a huge challenge to the radiologist in identifying and recognizing the cancer region due to the large volume of images as a result of increasing cancer cases (Teshale et al., 2019). As a result, detection accuracy rate tends to decrease. Giant strides have been made toward improving classification accuracy through application of machine learning algorithms like support vector machines, K-Nearest Neighbor and decision trees. The approach tried to automate the process by using a separate feature  extraction  method  which  is  followed  by  classification  algorithm,  in  theory,  this approach slightly increase the accuracy of screening mammogram (Falconi et al., 2019). However,  these  newly  developed  algorithms  could  not  effectively  operate  on  a  raw  data because  it  required  important  features  like  edges,  shapes,  textures  and  colors  to  be  first extracted from the image by human before being fed into the models (Wang et al., 2020). Most recent approach involves the use of deep learning techniques it can automatically learn the features from the mammograms directly. It has greatly increased the accuracy of mammogram screening for early signs of breast cancer detection. However, the major limiting factors of these techniques are that it requires high computing resources, large amount of data to train, making them hard to optimize. Hence, the research seeks to use a modified Convolution Neutral Network Technique for breast cancer classification.

1.3      Aim and Objectives of the Research

The aim of this study is to improve on classification accuracy in determining the likelihood of  cancerous  areas  from  mammogram  images  using  modified  Convolution  Neutral Network Technique. To achieve this, the following objectives are to:

I.      Acquire dataset from Mammographic Image Analysis Society (MIAS) database website and process the dataset using MATLAB

II.      Design a deep learning model architecture using modified Convolution Neutral Network for breast cancer classification

III.      Evaluate the performance of the model using system accuracy, sensitivity, specificity and F1-score

IV.     Compare the research work with existing research in literature using deep learning and MIAS breast cancer dataset

1.4      Scope of the Study

The scope is limited to breast cancer classification using deep learning in MATLAB environment which does not require physical development.

1.5     Justification for the Study

Breast cancer is one of the most common forms of cancer amongst women with statistics indicating that 1 in 7 females will be diagnosed with breast cancer in their lifetime.

Early detection of breast cancer through screening tests such as mammograms is an efficient way to maximize patient survival rate by treating the disease. However, no matter the expertise of radiologists examining mammograms, external factors such as fatigue, human error and interpretation error need to be minimized (Hepsag et al., 2017). Error- prone mammogram interpretations by radiologists can lead to decisions that can ultimately harm the patients. To that end, using Computer-Assisted Detection (CAD) software can help minimize the number of wrong interpretations and increase the accuracy of mammography screening is justifiable as an area of interest.

1.6     Thesis Outline

The remaining part of the thesis is organized as follows; Chapter two reviews Breast cancer, Early Breast Cancer detection systems, Machine Learning applications to medical image analysis, and Deep Learning applications to medical image analysis and also related works of literature on Breast cancer classification using Convolutional Neutral Networks. Chapter three presents  the research  system  model,  and  methodology of Breast  cancer classification using Convolution Neutral Network techniques. Chapter four presents and discusses  the MATLAB  simulation  results  of the research  and  chapter  five states  the conclusion and recommendations of the research.



This material content is developed to serve as a GUIDE for students to conduct academic research


BREAST CANCER CLASSIFICATION USING MODIFIED CONVOLUTION NEUTRAL NETWORK TECHNIQUE

NOT THE TOPIC YOU ARE LOOKING FOR?



Project 4Topics Support Team Are Always (24/7) Online To Help You With Your Project

Chat Us on WhatsApp »  09132600555

DO YOU NEED CLARIFICATION? CALL OUR HELP DESK:

   09132600555 (Country Code: +234)
 
YOU CAN REACH OUR SUPPORT TEAM VIA MAIL: [email protected]


Related Project Topics :

Choose Project Department