Master's Theses or Doctor of Nursing Practice

Department

Computer Science

Degree Name

Master of Science (MS)

Abstract

Accurate classification of blood cell types is a critical task in automated hematological analysis. This study presents a comparative evaluation of three deep learning architectures, ResNet18, MiniVGG, and YOLOv8, for five-class blood cell image classification. To ensure a fair comparison, all models were trained under standardized conditions, including a consistent 90:10 training–validation split, controlled dataset size, and fixed training epochs. ResNet18 was trained to establish a baseline using residual learning. MiniVGG employed a compact VGG-inspired design with regularization to balance efficiency and accuracy, while YOLOv8 leveraged a lightweight, pretrained classification backbone with integrated data augmentation. Experimental results demonstrate a clear progression in performance across the models. ResNet18 achieved a validation accuracy of 0.742, with misclassifications primarily occurring between visually similar cell types. MiniVGG significantly improved performance, reaching 0.962 accuracy and an average F1-score of 0.961, indicating strong and consistent predictions across all classes. YOLOv8 delivered the best results, achieving near-perfect classification with 0.996 accuracy and identical precision, recall, and F1-scores of 0.996. These findings highlight the effectiveness of modern lightweight architectures and pretrained models for medical image classification, with YOLOv8 emerging as a highly robust and reliable solution for blood cell analysis.

Keywords

Artificial intelligence, Convolutional neural networks, Multiclass classification, Transfer learning, Medical imaging diagnostics

Advisor

Dr. Anas Hourani

Date of Award

Spring 2026

Document Type

Thesis

Rights

© The Author

Download


Share

COinS