Methods of AI in Classification of Antepartum Heart Rate

Bc. Mohamed Darwish

Diplomová práce

Abstract:

Continuous monitoring of the fetal heart rate (FHR) signal has been widely used to allow obstetricians to obtain detailed physiological information about newborns. The objective of this thesis is to examine a novel approach of analyzing FHR signals with the help of AI and propose a framework that can help obstetricians make better decisions. This work is composed of mainly two parts in which the first part showcases the use of a CNN model to help classify the FHR signals into normal and pathological categories while the second part discusses the potential of integrating the CNN model with the already existing medical infrastructure. We proposed transforming the signals into 2D images using wavelet transformation and feeding it into a deep CNN model. The model was able to achieve a specificity of 84% and sensitivity of 30%. We then discussed the potential of utilizing the model in the cloud to turn it into a reliable CAD system.

Keywords

Fetal Heart Rate, Signal Processing, Deep Learning, Wavelet Transform, Scalogram, Convolution Neural Network

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 3. 4. 2023

Obhajoba závěrečné práce

• Obhajoba proběhla 22. 5. 2023
• Vedoucí: Ing. Petr Hanzlík, Ph.D.
• Oponent: Petr Vlachynský, Ing.

Česká zemědělská univerzita v Praze

Provozně ekonomická fakulta

Magisterský studijní program:
Informatics