DEEP LEARNING DETECTION OF COVID-19 PNEUMONIA USING CHEST COMPUTED TOMOGRAPHY IMAGES

MOK, ZHUANG DI (2022) DEEP LEARNING DETECTION OF COVID-19 PNEUMONIA USING CHEST COMPUTED TOMOGRAPHY IMAGES. [Final Year Project Report] (Unpublished)

PDF (Please get the password by email to repository@unimas.my , or call ext: 082-583914/3973/3933) Mok Zhuang Di.pdf Restricted to Registered users only Download (2MB)

Abstract

The Coronavirus disease 2019 also known as COVID-19 is a type of contagious disease that led to a worldwide on-going pandemic that was first identified in Wuhan, China in December 2019. In the period of rapid spread of the pandemic, several methods had been invented and tested in the field of image processing to resolve the time consuming and labour intensive to detect COVID-19 in early stages. In this project, an Artificial Intelligence (AI) deep learning image processing system is introduced to classify and detect COVID-19 with chest computed tomography (CT) scan. A total of 3,777 CT scan images are collected from reliable sources to train the deep learning neural network to perform classification. The deep learning neural network and image processing were processed by using MATLAB R2020a software with in-app toolbox extensions. The architecture of the whole project is divided into a few stages including image filtration and selection, image pre-processing, setting up training and testing sets, constructing and train deep learning neural networks, and lastly performance benchmarking. Several designs of pre-trained deep learning neural network will be tested and brought to performance comparison. The highest performing neural network is hyperparameter-tuned VGG-16 with validation accuracy of 99.7% and testing accuracy of 99.2%.

Actions (For repository members only: login required)

View Item