You are on page 1of 4

1.

INTRODUCTION

1.1 The Background of the Study

Cancer is one type of disease that has a high increase in the number of
cases in the world. Based on data obtained from WHO (World Health
Organization), in 2018 there were 18.1 million new cases with a mortality rate of
9.6 million deaths, where 1 in 5 men and 1 in 6 women in the world experience
cancer. The data also stated that 1 in 8 men and 1 in 11 women die from cancer.
The rate incidence of cancer in Indonesia is (136.2 / 100,000 population) which
ranks 8th in Southeast Asia and 23rd in Asia.
Histopathology or the study of tissue, has a major contribution in
determining the prognosis of cancer at the tissue level. Quantitative analysis of
histopathological images is used to determine the treatment that should be
performed. That way early detection can be done properly and can significantly
improve patient survival.
However, histopathological analysis is a relatively tiring process because
of the complexity of the structure in different tissues. Pathologists need a lot of
expertise and skills to do analysis. Qualitative examination of the large amounts
of data presented in the picture is a tiring task and requires professional working
hours. In addition, visual inspection of images by experts can lead to inaccurate
results because of their subjective nature.
With the development of machine learning algorithms (machine learning)
and computer vision, computerized histopathological analysis can be augmented
by image analysis that utilizes this knowledge to facilitate quantitative and
qualitative examinations. That way, the results of the analysis can have a better
level of accuracy.

1
Deep Learning or often known as Deep Structured Learning or
Hierarchical Learning is one branch of machine learning that consists of high
level abstraction modeling algorithms in data using a set of data non-linear
transformation functions arranged in layers and depth.
According to this, a program can be made that is used to analyze medical
digital images to show the results of histopathological analysis done manually, the
title of research study conducted was "Implementation of Convolutional Neural
Network on Histopathologic Image Detection for Cancer Metastasis Conjecture".

1.2 The Identification of the Problems


Based on the description from the background described on the
background of the study, some problems that can be formulated include:
1. How to classify the following histopathological image?
2. How can the results of the classification be used to detect cancer suspicions?
3. How to do an accuracy test on the results of detections?
1.3 The Limitation of the Study
Based on the background above, as for some of the limitations of the
problem in this study include:
1. The data used is secondary data obtained through a dataset containing binary
classification images containing 200,025 images with low resolution (96 x
96px) from the lymph node tissue extracted from digital histopathological
scanning.
2. The ratio of the number of image labels used for training and validation
testing is 50:50.
3. The program is made with the Python 3 programming language and
supporting libraries for computer vision programming.
4. The program is used to classify images that have cancer metastasis and normal
images. Then detection is made on the tested data.

2
1.4 The Objective of the Study
The general purpose of this paper is to make a program to detect cancer
suspects through histopathological imagery that can be used to support the
decision of a pathologist in making a diagnosis of cancer in patients. While the
specific purpose of this study is:
1. Generate program that can classifies and detect cancer conjecture.
2. Generate a program with a good degree of accuracy to help pathologists
classify data and detect cancer allegations.
1.5 Research Methods
In this writing, there are several stages used in conducting research, which
are as follows:
1. Research planning
At this stage, background and problem boundaries are determined. After that,
a literature study is conducted to collect methods that can be used to classify
and detect cancer suspicions.
2. Data Collection Phase
After the planning is done, a collection of datasets containing digital images of
tissue in the lymph glands are collected indirectly through the Kaggle site.
3. Analysis
At this stage the analysis is carried out there are methods obtained from
library studies with the best level of ease of implementation, performance, and
accuracy.
4. Generate the Program
After the data has been collected and program planning has been made, a
model is defined and training data is done using the Python 3 programming
language. After the model has been created then the model is stored.
5. Program Testing
At this stage the program is tested with data that has been collected.
6. Accuracy Test

3
After testing the program, an accuracy test is performed to obtain accurate
detection results.

1.6 Writing Systematics


In the writing systematics, a general description of the presentation of the
problem is explained and divided into five chapters with the following details:
1. INTRODUCTION
In this chapter the background of the problem, problem boundaries, problem
formulation, the purpose of the writing, the research methods and the
systematic writing are explained.

2. THEORITICAL REVIEW
In this chapter contains the theoretical foundation needed in making research.

3. ANALYSIS AND DESGIN


This chapter consists of the details of the methods used in solving the
problems presented. Problem solving is done by designing program that can
process data, classifies and detect the problems presented in the study.

4. RESULTS AND DISCUSSION


In this chapter we will describe the stages of making a program, the process of
training data, and the results of testing of the program being made.

5. CONCLUSION AND SUGGESTION


The closing chapter contains a description of the conclusions obtained during
the writing and suggestions for the development of what has been presented in
the results of the study.

You might also like