Tools
Tools
Meshram, Mrunal Dhanraj (2024) Hybrid Skin Disease Diagnosis Using StyleGAN2 and EfficientNet. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (3MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (1MB) | Preview |
Abstract
Automated skin disease detection is a transformative application of machine learning in healthcare, addressing the challenges of early diagnosis and effective treatment planning. This study introduces a hybrid framework that integrates StyleGAN2 and UNet for synthetic data generation and EfficientNet-B5 for classification, tackling issues such as data scarcity, class imbalance, and variability in dermatological datasets. The curated dataset includes three skin disease categories eczema, psoriasis, and fungal infections chosen for its diagnostic complexity and clinical relevance. Through synthetic data augmentation, the framework achieved significant improvements in classification performance, with accuracy increasing from 68.5% to 82.3% and F1-score rising from 0.72 to 0.85. Synthetic images generated using the StyleGAN2-UNet hybrid model exhibited a low Fr´echet Inception Distance (FID) score of 18.7, validating their quality and utility. Evaluation metrics such as precision, recall, and F1-scores were supplemented by visual tools like confusion matrices, ROC curves, and Class Activation Maps (CAMs), ensuring both reliability and interpretability. This study contributes a scalable, robust, and interpretable solution for automated skin disease detection, with the potential for broader applications in medical diagnostics.
| Item Type: | Thesis (Masters) |
|---|---|
| Supervisors: | Name Email Shahid, Abdul UNSPECIFIED |
| Uncontrolled Keywords: | Skin Disease Detection; Synthetic Data Generation; StyleGAN2; UNet; EfficientNet-B5; Class Imbalance; Medical Imaging |
| Subjects: | Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science R Medicine > RL Dermatology R Medicine > Healthcare Industry Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning |
| Divisions: | School of Computing > Master of Science in Artificial Intelligence |
| Depositing User: | Ciara O'Brien |
| Date Deposited: | 20 Jun 2025 08:44 |
| Last Modified: | 20 Jun 2025 08:44 |
| URI: | https://norma.ncirl.ie/id/eprint/7955 |
Actions (login required)
 |
View Item |