Please use this identifier to cite or link to this item:
Title: Automatic Covid-19 Chest CT Image Classification
Authors: LOK, CHON WENG(陸晉榮)
Department: Department of Computer and Information Science
Faculty: Faculty of Science and Technology
Issue Date: 2021
Citation: Lok, C. W. (2021). Automatic Covid-19 Chest CT Image Classification (PPO) Algorithm in Carla (OAPS)). Retrieved from University of Macau, Outstanding Academic Papers by Students Repository.
Abstract: Background and Purpose: COVID-19 pandemic as a Public Health Emergency of International Concern in January 2020, we can know that the severity in that moment. COVID-19 pandemic influences many areas from world economics to personal health. Base on COVID-19 pandemic can be distinguish by CT chest image, our project is we wish to building a high-performance COVID-19 chest image classifier. Material and Method: Our classifier make up by CT chest image in 416 COVID-19 patients, 412 common pneumonia patients, and 270 health patients. We use the neural network as the classifier of CT image. In order to make the result more dependable and visualization, we are going to build the website user interface, the diagnostic result of the chest CT image and network model will show on website. The website structure as following. 1. Introduction of the project (including Motivation, Technical Challenges, and Project Objectives) 2. Related work 3. Basic ideas of the platform(s), solution(s), and contribution(s) Results: Our model can accurately identify three kinds of patients, common pneumonia patients, COVID-19 patients, and health patients. Following accuracy is 86.67%, 94.59%, and 97.37%
Course: Bachelor of Science in Computer Science
Instructor: Prof. Yibo Bob ZHANG
Programme: Bachelor of Science in Computer Science
Appears in Collections:FST OAPS 2021

Files in This Item:
File Description SizeFormat 
OAPS_2021_FST_DB726058_Lok ChonWeng_Automatic Covid-19 Chest CT Image Classification.pdf39.66 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.