Sample Publications:

AI-Powered Nailfold Capillary Analysis

Adel Elmaghraby, Mona Ebadi Jalal, Omar S. Emam, Cristián Castillo-Olea, and Begoña García-Zapirain introduce a deep learning model using EfficientNet-B0 with cascade transfer learning for detecting abnormalities in nailfold capillary images. Their model achieved perfect classification accuracy (1.00), significantly outperforming traditional methods. This AI-driven approach enhances early disease detection and improves diagnostic efficiency in conditions like diabetes and cardiovascular diseases. The study highlights the potential of AI in medical imaging and aims for further validation on larger datasets for clinical applications.

Bias-Resistant Corpus for Emotion Recognition

The paper presents a gender-bias-resistant super corpus designed to improve fairness in Speech Emotion Recognition (SER). By merging and balancing multiple existing datasets, the authors create a comprehensive resource with equal male and female representation. Deep learning models trained on this corpus show improved emotion recognition accuracy across genders, highlighting the importance of addressing bias in SER systems.

PSO-Based Biomedical Image Registration

Adel S. Elmaghraby, alongside Mark P. Wachowiak, Renata Smolíková, Yufeng Zheng, and Jacek M. Zurada, introduces a groundbreaking image registration method using Particle Swarm Optimization (PSO). This hybrid approach leverages user guidance and evolutionary strategies to enhance accuracy and efficiency in aligning 2D and 3D biomedical images. The technique significantly surpasses traditional methods, showcasing its potential for advancing diagnostics, treatment planning, and image-guided interventions.

Deep learning role in early diagnosis of prostate cancer

The paper investigates how deep learning can be utilized for the early detection of prostate cancer, enhancing diagnostic accuracy and efficiency. Dr. Adel Elmaghraby played a key role in this research, focusing on integrating AI techniques into medical imaging to support early diagnosis and better patient outcomes.