Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/112629
Title: | Ekstraksi Fitur Klinis Citra CT Scan Penderita COVID-19 Menggunakan Metode Texture Analysis |
Authors: | Herdiyeni, Yeni Wulandari Astuti, Annisa Widia |
Issue Date: | 2022 |
Publisher: | IPB University |
Abstract: | Pada akhir tahun 2019, muncul penyakit COVID-19 yang disebabkan oleh
coronavirus. Penyakit ini kemudian dinyatakan sebagai pandemi oleh WHO pada
akhir bulan Januari 2020. Metode standar yang disarankan oleh panduan diagnosis
COVID-19 adalah menggunakan RT-PCR dan analisis citra. Namun, kedua metode
ini masih belum cukup optimal dikarenakan waktu yang dibutuhkan untuk
melakukan identifikasi COVID-19 cukup lama, yaitu 4-6 jam menggunakan RTPCR dan 21,5 menit menggunakan analisis citra. Selain itu, sensitivitas yang
dihasilkan menggunakan metode RT-PCR masih cukup rendah, yaitu 71%.
Penelitian ini bertujuan untuk menerapkan algoritme texture analysis untuk
melakukan identifikasi COVID-19 berbasiskan citra computed tomography. Fitur
klinis yang didapatkan pada texture analysis kemudian diklasifikasikan
menggunakan metode support vector machine. Hasil klasifikasi yang didapatkan
cukup baik dengan nilai precision 86%, recall (sensitivitas) 99%, dan accuracy
91%. Diharapkan metode ini dapat membantu tenaga medis dalam melakukan
proses identifikasi COVID-19. At the end of 2019, the COVID-19 disease caused by the coronavirus emerged. The disease was later declared a pandemic by WHO at the end of January 2020. The standard method recommended by the COVID-19 diagnosis guide is to use RT-PCR and image analysis. However, these two methods are still not optimal enough because the time needed to identify COVID-19 is quite long, i.e. 4-6 hours using RT-PCR and 21.5 minutes using image analysis. In addition, the sensitivity produced using the RT-PCR method is still quite low, i.e. 71%. This study aims to apply a texture analysis algorithm to identify COVID-19 based on computed tomography images. The clinical features obtained in the texture analysis are then classified using the support vector machine method. The classification results obtained are quite good with 86% precision, 99% recall (sensitivity), and 91% accuracy. It is hoped that this method can help medical personnel in carrying out the COVID-19 identification process. |
URI: | http://repository.ipb.ac.id/handle/123456789/112629 |
Appears in Collections: | UT - Computer Science |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
G64170051 Annisa Widia Astuti.pdf Restricted Access | Fullteks | 3.28 MB | Adobe PDF | View/Open |
G64170051 Annisa Widia Astuti-1-12.pdf Restricted Access | Cover | 1.77 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.