Klasifikasi Non Small Cell Lung Cancer (NSCLC) Dengan Menggunakan Convolutional Neural Network (CNN)

Darmawan, Bunga Mastiti (2021) Klasifikasi Non Small Cell Lung Cancer (NSCLC) Dengan Menggunakan Convolutional Neural Network (CNN). Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

Text 01111740000050- Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (3MB) | Request a copy

Abstract

Kanker paru-paru mempunyai dua jenis berdasarkan cara penanganannya yaitu Small Cell Lung Cancer (SCLC) dan Non Small Cell Lung Cancer (NSCLC). Kanker paru SCLC biasanya berada pada pasien yang memiliki riwayat perokok berat dengan penyebaran yang lebih cepat dibanding dengan NSCLC. Namun sekitar 8085% dari seluruh kasus kanker paru adalah jenis NSCLC yang banyak menyerang pria maupun wanita. Penelitian ini mempunyai tujuan untuk mengklasifikasikan NSCLC ke dalam paru-paru Normal, Adenokarsinoma, Karsinoma sel skuamosa, dan Karsinoma sel besar, serta untuk mengetahui perbandingan hasil klasifikasi menggunakan arsitektur VGG19 dan ResNet50. Data yang digunakan yaitu berupa Citra CT Scan pada masing-masing klasifikasi dengan jumlah total gambar sebanyak 1000 gambar. Proses pengklasifikasian dilakukan kedalam 3 proses yaitu preprocessing, klasifikasi, dan validasi. Pada tahap preprocessing data yang ada dilakukan proses resize dan grayscale untuk menyeragamkan semua gambar yang akan di input. Didapati hasil paling optimal pada arsitektur ResNet50 yaitu dengan hasil akurasi sebesar 99,87% pada data training, data test sebesar 98,35%, dan pada data validasi sebesar 96%. Dengan hasil klasifikasi terbaik pada kelas Normal yaitu didapati hasil presisi, sensitivitas, F1 Score, dan spesifisitas pada data validasi berturut-turut sebesar 100%, 100%, 100%, dan 100%.
======================================================================================================
Lung cancer has two types based on how it is treated, namely Small Cell Lung Cancer (SCLC) and Non-Small Cell Lung Cancer (NSCLC). SCLC lung cancer usually occurs in patients with a history of heavy smoking and spreads more rapidly than NSCLC. However, about 8085% of all lung cancer cases are NSCLC types that mostly attack men and women. The purpose of this study was to classify NSCLC into the normal lung, adenocarcinoma, squamous cell carcinoma, and large cell carcinoma and compare the classification results of the architecture VGG19 and ResNet50. The data using is in the form of CT Scan images in each classification with 1000 images. The classification process is carried out into three processes, namely preprocessing, classification, and validation. At the preprocessing stage of the existing data, the resizing and grayscale process is carried out to uniform all the images that will be input. In the VGG19 and ResNet50 sections, a transfer learning process is carried out to train new parameters for classifying NSCLC types. The most optimal result was found in the ResNet50 architecture, with an accuracy of 99.87% on training data, 98.35% on test data, and 96% on data validation. With the best classification results in the normal class, it was found that the results of precision, sensitivity, F1 Score, and specificity of the validation data were 100%, 100%, 100%, and 100%, respectively.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Deep Learning, Classification, Lungs, Deep Learning, Klasifikasi, Paru paru.
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QC Physics Q Science > QR Microbiology > QR 201.T84 Tumors. Cancer
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Physics > 45201-(S1) Undergraduate Thesis
Depositing User:	Bunga Mastiti Darmawan
Date Deposited:	10 Aug 2021 22:13
Last Modified:	10 Aug 2021 22:13
URI:	http://repository.its.ac.id/id/eprint/85342

Actions (login required)

View Item