Deteksi Pneumonia Pada Anak-Anak Dari Citra X-Ray Berbasis Convolutional Neural Network

Fauziyah, Alif Hilmi (2020) Deteksi Pneumonia Pada Anak-Anak Dari Citra X-Ray Berbasis Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 07311640000013-Undergraduate_Thesis.pdf]
Preview
Text
07311640000013-Undergraduate_Thesis.pdf

Download (7MB) | Preview

Abstract

Deteksi suatu penyakit melalui suatu citra X-Ray masih bergantung pada diagnosis
tenaga medis. Deteksi secara manual berdasarkan pengamatan visual dari tenaga
medis masih banyak dilakukan untuk membaca hasil foto X-Ray. Cara manual ini
memiliki kelemahan yaitu adanya perbedaan interpretasi antar pengamat sehingga
dapat terjadi kesalahan dalam diagnosis suatu penyakit. Serta membutuhkan waktu
yang cukup lama saat mendiagnosis secara manual. Pada penelitian ini dibuat suatu
sistem untuk mendeteksi pneumonia pada anak-anak dari hasil foto X-ray dengan
menggunakan metode Convolutional Neural Network (CNN). Tahap pemrosesan
awal juga dilakukan untuk meningkatkan kualitas citra. Metode median filter dan
contrast limited adaptive histogram equalization (CLAHE) diterapkan untuk image
enhancement sebelum citra memasuki proses CNN. Pada CNN terdapat jaringan
dengan 2 lapisan utama, yaitu feature learning yang terdiri dari convolution layer
dan pooling layer, serta lapisan klasifikasi yang terdiri dari fully connected layer.
Pada fully connected layer akan diberlakukan fungsi aktivasi sigmoid untuk
mengklasifikasikan citra ke dalam 2 kelas, yaitu normal dan pneumonia. Hasil akhir
yang didapatkan dalam tugas akhir ini adalah performa model yang cukup baik
dengan akurasi yang didapatkan yaitu sebesar 96,06 % untuk akurasi validasi dan
akurasi pelatihan sebesar 96,18 %. Confusion matrix juga digunakan untuk
mengukur kinerja sistem. Berdasarkan confusion matrix perhitungan presisi, recall,
dan spesifisitas yang didapatkan adalah 97,83 %, 97,12 %, dan 92,27 %. Kurva
ROC yang dihasilkan juga menunjukkan bahwa nilai Area Under Curve (AUC)
sebesar 0,95 dengan grafik yang menjauhi garis baseline.
==================================================================================================================
Detection of the disease through X-Ray images is still according to the diagnosis of
medical personnel. Manual detection of visual reading from medical personnel is
still mostly done to read X-Ray images. This manual method has the disadvantage
that there are differences in interpretation between observers so that errors can occur
in the diagnosis of a disease. And requires quite a long time when diagnosing
manually. In this study a system was made to test pneumonia in children from Xray images using the Convolutional Neural Network (CNN) method. The initial
maintenance phase is also carried out to improve image quality. The median filter
and contrast limited adaptive histogram equalization (CLAHE) method is applied
for image enhancement before the process of CNN. CNN consists of 2 main layers,
namely the feature learning feature consists of a convolution layer and a pooling
layer, as well as a classification layer consist of a fully connected layer. In the fully
connected layer the sigmoid activation function will be applied to classify the image
into 2 classes, there is normal and pneumonia. The final results obtained in this
thesis is a good performance model with an accuracy is 96.06% for validation
accuracy and training accuracy is 96.18%. The confusion matrix is also used to
measure system performance. Based on the confusion matrix the calculation of
precision, recall, and specificity obtained were 97.83%, 97.12%, and 92.27%. The
resulting ROC curve also shows an Area Under Curve (AUC) value of 0.95 with a
graph that moves away from the baseline.

Item Type: Thesis (Other)
Uncontrolled Keywords: Pneumonia, Convolutional Neural Network, CLAHE, Median filter Pneumonia, Convolutional Neural Network, CLAHE, Median filter
Subjects: Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
R Medicine > RJ Pediatrics
T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques
T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7882.P3 Pattern recognition systems
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Biomedical Engineering > 11410-(S1) Undergraduate Thesis
Depositing User: Fauziyah Alif Hilmi
Date Deposited: 26 Aug 2020 05:58
Last Modified: 16 Nov 2023 07:09
URI: http://repository.its.ac.id/id/eprint/80480

Actions (login required)

View Item View Item