Prediction of Covid-19 risk of a person by analyzing computed tomography images using convolutional neural networks

Date

2024-1-24

Author

Topçu, Kaan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

138
views

57
downloads

Cite This

In this thesis, 4 main research questions are answered to evaluate the performance of convolutional neural networks (CNN) in predicting Covid-19 risk by using computed tomography (CT) images. The CT images by Yang et al., 2020 are utilized for this study. It contains 349 CT images labeled as being positive for Covid-19 from 216 patient and 397 CT images that are negative. Different CNNs like VGG-16, ResNet- 18, ResNet-50, DenseNet-121, DenseNet-169, EfficientNet-B0, and EfficientNet-B1 are experimented and evaluated accordingly. The first research question investigates the performance of the CNNs without pretraining them. The second one evaluates the effect of transfer learning for each CNN. The third research question studies the impact of source dataset’s domain used for transfer learning. Finally, whether the performance of the networks can be maintained or improved by training the networks partially is analyzed. iv All four research questions are evaluated by comparing the accuracy, F1-score and AUC values.

Subject Keywords

Convolutional neural network, Covid-19, Image classification, Computed tomography

URI

https://hdl.handle.net/11511/108466

Collections

Graduate School of Informatics, Thesis