Improving Medical Diagnosis with Deep Learning Models in Two Medical Image Analyses: Comparative Performance and Future Horizons
Albanian Journal of Natural and Technical Sciences · 30.12.2024 · 29 (2) · 93-110
DOI: 10.67537/ajnts2902018 ↗Kurz gefasst
Fachartikel in der Zeitschrift Albanian Journal of Natural and Technical Sciences, online erschienen am 30.12.2024. Verfasst von Bekir Karlik und 1 weiteren Autor:innen. Die Arbeit nennt 23 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This study explores the application of deep learning (DL) techniques in medical image analysis, with a focus on disease identification using neural network models applied to various imaging modalities. The exponential growth of medical imaging data, coupled with the urgent need for accurate diagnoses, has driven the adoption of convolutional neural networks (CNNs) and transfer learning strategies in healthcare. While these models have significantly improved diagnostic accuracy, they still face challenges such as data scarcity for rare diseases, class imbalance, and limited generalizability to real-world clinical settings. Comparative experiments using MRI and chest X-ray datasets highlight architecture-specific performance and provide insights into training dynamics, overfitting tendencies, and model robustness. The study evaluates the performance of leading DL models—ResNet50, Xception, VGG16, InceptionV3, and EfficientNetB0—based on accuracy and loss metrics. The results indicate that the ResNet50 model is best suited for the pneumonia dataset, whereas the Xception model performs more effectively on the brain tumor dataset. The study concludes by proposing future research directions, including the integration of self-supervised learning, multimodal data fusion, and federated learning, with the aim of enhancing the scalability, robustness, and fairness of DL-based medical imaging solutions.
Zitieren
APA
Karlik, B., & Tafa, B. (2024). Improving Medical Diagnosis with Deep Learning Models in Two Medical Image Analyses: Comparative Performance and Future Horizons. Albanian Journal of Natural and Technical Sciences, 29(2), 93-110. https://doi.org/10.67537/ajnts2902018
Vancouver
Karlik B, Tafa B. Improving Medical Diagnosis with Deep Learning Models in Two Medical Image Analyses: Comparative Performance and Future Horizons. Albanian Journal of Natural and Technical Sciences. 2024;29(2):93-110. doi:10.67537/ajnts2902018
BibTeX
@article{Karlik2024,
author = {Karlik, Bekir and Tafa, Bora},
title = {Improving Medical Diagnosis with Deep Learning Models in Two Medical Image Analyses: Comparative Performance and Future Horizons},
journal = {Albanian Journal of Natural and Technical Sciences},
year = {2024},
volume = {29},
number = {2},
pages = {93--110},
doi = {10.67537/ajnts2902018}
}
EUVIMED