A DEEP LEARNING APPROACH TO DETECT LUNG CANCER USING ALEXNET AND KNN

  • : Ms Word, Ms Word Format
  • : 50 Pages
  • : ₦5,000
  • : 1-5 Chapters
  •  
  • Click to DOWNLOAD Materials

A DEEP LEARNING APPROACH TO DETECT LUNG CANCER USING ALEXNET AND KNN

ABSTRACT

In the race of all cancerous diseases, lung cancer is in first place. Every year lots of people died because of cancer and lung cancer is playing the leading role among them. In the year 2018, 9.6 million people died because of cancer whereas 1.76 million death occurred due to lung cancer. In this study, we experiment with a deep learning model with a kNN classifier to extend the success rate in diagnosing lung cancer. The dataset used in this study is a publicly accessible resource SPIE-AAPM. We used data augmentation on the training dataset to expand the dataset and convolutional neural network (CNN) to extract the related features. Extracted features from CNN are used as input to the kNN classifier with cross-validation. The experiment hit an accuracy of 90% by predicting the dataset with the help of selected features and the kNN classifier.

CHAPTER 1

INTRODUCTION

1.1 Background

Cancer is an unintentional enhancement of cells that could outspread in an uncontrolled manner, and sometimes it spreads. Among hundreds of cancer diseases, lung cancer has turned out one of the most common cancer in the world [18] . The beginning of lung cancer happens from the unusual development of lung cells. The lung absorbs oxygen when humans breathe in, and it discharges carbon dioxide when it breathes out. So, the lung of the human body is a very important and sensitive organ for humans. A very vast amount of people die every year because of lung cancer [1] . According to the Global Cancer Statistics, lung cancer and breast cancer had the same amount of new cases detected (lung cancer 11.6% of total cases and breast cancer also 11.6% of total cases). But, in the death ratio lung cancer is three times greater than breast cancer (lung cancer is 18.4% of total deaths and breast cancer is 6.6% of total deaths). A total of 2.1 million people were affected and 1.76 million people died because of lung cancer worldwide in 2018 [19] . The survival rate is too low because of late diagnosis. Hence, an early and faultless diagnosis of lung cancer may reduce the number of deaths.

The use of artificial intelligence in medical departments is increasing day by day at a massive rate. In the biomedical field, AI-based solutions are having great success. In recent years, deep learning is proven as a more effective technique than machine learning because of the power of relative feature extraction in the biomedical field.

1.3 Motivation for the Research

Every year lots of people are beaten by lung cancer and passed away in the afterlife. In 2018 2.1 million people were affected by lung cancer and from there 1.76 million people stop their journey and passed away in the afterlife. Due to late diagnosis only, a few people can survive it. Early diagnosis may reduce the number of death and increase the survival rate. There are various models were developed to identify precisely this disease earlier.

1.3 Problem Statement

It is really difficult and time-consuming for a field expert to diagnose lung cancer from CT scan images manually. Sometimes they cannot even diagnose cancer precisely. As a result, the patient has to face a confusing situation. The manual process of diagnosing lung cancer is also quite expensive.

1.4 Research Questions

According to the background, motivation, and problem statement, the following questions are raised:

  • Can the model used in this study detect lung cancer accurately?
  • Can the technique provide better results compared to the existing solutions?
  • Is this model more effective than others when it comes to comparison?

1.5 Research Objectives

  • To propose a pre-trained model which can perform with better perfection.
  • To produce a significant model which can help to detect lung cancer precisely.
  • To provide a better solution that can provide better results among the existing solutions.

1.6 Research Scope

The purpose of this study is to detect only lung cancer in CT scan images. Any other diseases or any other kind of cancer cannot be diagnosed using this system. The input images must be CT scan images of the lung, not other images like CT images of the abdomen, CT images of the brain, x-ray of the chest, etc. This implementation detects binary 0 and 1, whether the image is cancerous or non-cancerous. The system cannot detect the stages of lung cancer. The system can be used in a hospital where lung cancer patients exist or come for diagnosing lung cancer.

A DEEP LEARNING APPROACH TO DETECT LUNG CANCER USING ALEXNET AND KNN. GET MORE COMPUTER SCIENCE PROJECT TOPICS AND MATERIALS

 

Sharing is caring!

Leave a Reply