Kautsar, Ahmad ‘Izzan and Bayu Rahayudi, S.T., M.M. and Dr. Lailil Muflikhah, S.Kom., M.Sc (2024) Analisis Performa Model Convolutional Neural Network Dalam Mendeteksi Penyakit Pneumonia. Sarjana thesis, Universitas Brawijaya.
Abstract
Pneumonia merupakan infeksi paru-paru yang signifikan dan mematikan, dengan peningkatan kasus yang dipicu oleh polusi udara akibat emisi kendaraan. Saat ini, deteksi pneumonia umumnya dilakukan melalui analisis manual citra X-ray oleh tenaga medis. Namun, dengan kemajuan teknologi, penelitian terkait deteksi digital pneumonia menggunakan citra X-ray semakin berkembang. Convolutional Neural Networks (CNN) yang menjadi bagian dari Deep learning telah terbukti efektif dalam tugas klasifikasi gambar, termasuk dalam penelitian ini. Penelitian ini melakukan analisis model CNN melalui beberapa skenario percobaan, yang melibatkan variabel seperti mode warna, perbandingan data, dan penerapan augmentasi. Menggunakan 2000 data yang diambil secara acak dan dilakukan undersampling untuk meratakan persebaran, hasil penelitian menunjukkan bahwa skenario percobaan dengan perbandingan data 90:10, mode warna grayscale, dan tanpa augmentasi menghasilkan performa terbaik dengan akurasi 95%, presisi 95%, dan recall 95%.
English Abstract
Pneumonia is a significant and deadly lung infection, with rising cases triggered by air pollution from vehicle emissions. Currently, pneumonia detection is typically conducted through manual X-ray image analysis by medical personnel. However, with technological advancements, research on digital pneumonia detection using X-ray images has increasingly developing. Convolutional Neural Networks (CNN), a part of deep learning, have proven effective in image classification tasks, including in this study. This research analyzes the CNN model through several experimental scenarios, involving variables such as color mode, data ratio, and the application of augmentation. Using 2000 randomly selected data and applied undersampling to balance distribution. The research results indicate that the experimental scenario with a 90:10 data ratio, grayscale color mode, and no augmentation produced the best performance with 95% accuracy, 95% precision, and 95% recall.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | 0524150318 |
| Uncontrolled Keywords: | Deep Learning, Convolutional Neural Network, Deteksi Pneumonia, Klasifikasi. Deep Learning, Convolutional Neural Network, Pneumonia Detection, Classification |
| Divisions: | Fakultas Ilmu Komputer > Sistem Informasi |
| Depositing User: | Sugeng Moelyono |
| Date Deposited: | 28 Nov 2024 04:51 |
| Last Modified: | 28 Nov 2024 04:51 |
| URI: | http://repository.ub.ac.id/id/eprint/229302 |
![[thumbnail of DALAM MASA EMBARGO]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text (DALAM MASA EMBARGO)
Ahmad 'Izzan Kautsar.pdf Restricted to Registered users only Download (4MB) |
Actions (login required)
 |
View Item |