CHAPTER ONE

INTRODUCTION

1. 1. Background of the Study

Early detection of skin cancer has the potential to reduce mortality and morbidity rate among cancer patients, today computer-aided decision systems have become important when evaluating and diagnosing medical images. For example, computer aided diagnosis (CAD) is part of the routine when detecting breast cancer on mammograms in the United States. CAD is also one of the major research subjects in medical imaging and diagnostic radiology. An accurate CAD system can be used for early detection of a disease and thereby allow for earlier treatment, which could save lives. For example, the ability to effectively treat and cure cancer is directly dependent on the ability to detect cancers at their earliest stages (Wulfkuhle et al., 2013).

Cancer is a collection of related diseases where diagnosis and treatment are of great interest due its widespread occurrence. In 2012, there were 14 million new cases of cancer and 8.2 million cancer-related deaths worldwide. This makes cancer one of the most common causes of death for humans. Skin cancer is the most common type of cancer and usually forms in skin that has been exposed to sunlight, however it can occur on any part of the body. Skin cancer begins in the epidermis (outer layer of the skin) and is therefore clearly visible. This means that a CAD has potential to use only images of the skin lesion, without any other information, in order to give a preliminary diagnosis (Pathan et al., 2018).

The most straightforward and effective solution to control the mortality rate for skin cancer is the timely diagnosis of skin cancer as the survival rate for melanoma patients in a five-year timespan is 99 percent when diagnosed and screened at the early stage (Waltz, 2017). Moreover, the most mundane skin cancer types Basal Cell Carcinoma (BCC) and Squamous Cell Carcinoma (SCC) are highly treatable when early diagnosed and treated adequately. Dermatologist primarily utilizes visual inspection to diagnose skin cancer, which is a challenging task considering the visual similarity among skin cancers. However, dermoscopy has been popular for the diagnosis of skin cancer recently considering the ability of dermoscopy to accurately visualize the skin lesions not discernible with the naked eye. Reports on the diagnostic accuracy of clinical dermatologists have claimed 80 percent diagnostic accuracy for a dermatologist with experience greater than ten years, whereas the dermatologists with experience of 3-5 years were able to achieve diagnostic accuracy of only 62 percent, the accuracy further dropped for less experienced dermatologists. The studies on Dermoscopy imply a need to develop an automated efficient, and robust system for the diagnosis of skin cancer since the fledgling dermatologists may deteriorate the diagnostic accuracy of skin lesions (Hameed et al., 2016).

1.2     Statement of the Problem

Skin cancer is an abnormal growth of skin cells. It generally develops in areas that are exposed to the sun, but it can also form in places that don’t normally get sun exposure, this has plagued many people around the globe, early detection of skin cancer can go a long way in saving the life of the cancer patient. Dermatologists are experts trained in the diagnosis of skin diseases among which, skin cancer is also one, however the accuracy of a dermatologist diagnosis, is based on his/her years of experience, this has been a major problem in the diagnosis of skin cancer, thus the initiation of our proposed study, in which Convolutional Neural Networks (CNN), a class of deep learning (DL), with end-to-end feature extraction and classification, promises a highly scalable and superior diagnosis accuracy on the classification of skin cancer using a mobile device.

1.3     Aim and Objectives of the Study

The aim of this study is to develop a skin cancer classification system using Deep Learning (DL). The objectives of the study are as follows;

1. To Design a convolution neural network (CNN) model for skin cancer classification.
2. To Implement the system in (i) using TensorFlow to train the classification model.
3. Deploy the model trained in (ii) on an android mobile app, to run on a mobile device.
4. To perform functional test on the system deployed in (iii).

1.4     Significance of the Study

This study will serve as an avenue that dermatologists or non-specialist individuals can use to carry out skin cancer classification, using their mobile device with better accuracy regardless of their years of experience.

The software is also relevant in the following ways:

1. Enable easy classification of skin cancer.
2. Reduce the cost of hiring an expert dermatologist to do the job of classifying skin cancer.
3. Reduce the time of carrying out skin cancer classification.

1.5     Scope of the Study

This research is limited to the classification of skin cancer by using computer vision and convolutional Neural Network on a mobile device.

1.6   Limitation of the Study

The researcher faced some limitations during the course of this research. Some of this limitations/constraint includes;

i.        Time constraint: Time factor was the major factors as the timeframe to submit the project was too short for a rigorous research.

ii.       Financial constraint:  Lack of adequate finance to fund the researcher as to the purchase of data to review related works online was another major setback.

1.7     Definition of Terms

1. Biopsy: This is an examination of tissue removed from a living body to discover the presence, cause, or extent of a disease.
2. Cancer Cells: These are cells that divide relentlessly, forming solid tumors or flooding the blood with abnormal cells.

• iii. Cancer: is a disease in which some of the body's cells grow uncontrollably and spread to other parts of the body. Cancer can start almost anywhere in the human body, which is made up of trillions of cells.

4. Convolutional Neural Network (CNN):   Is a class of Deep Neural Networks, most commonly applied to analyzing visual imagery.
5. Dataset:      A collection of related sets of information that is composed of separate elements but can be manipulated as a unit by a computer.
6. Disease Detection:    It is a process of determining which disease or condition explains a tree/persons symptoms and sign. It is most often referred to as diagnosis with the medical context.
7. Disease Diagnosis:   It is process of determining which disease or condition explains a tree/person’s symptom and signs.
8. Multilayer Perceptron: This refers to fully connected networks, that is, each neuron in one layer is connected to all neurons in the next layer.
9. Pathogen:   It is a bacterium, virus, or other microorganism that can cause disease.
10. Pathology: This is the study of the causes and effects of disease or injury.
11. Perceptron: This refers to a linear classifier (binary) use in supervised learning. It helps to classify the given input data.
12. Skin Cancer: the abnormal growth of skin cells — most often develops on skin exposed to the sun. But this common form of cancer can also occur on areas of your skin not ordinarily exposed to sunlight. There are three major types of skin cancer — basal cell carcinoma, squamous cell carcinoma and melanoma
13. Ultrasound: Sound or other vibrations having an ultrasonic frequency, particularly as used in medical imaging.