Leveraging Hybrid Deep Learning Approaches for Effective Sickle Cell Anemia Diagnosis from Microscopic Images

Sickle cell anaemia is a critical inherited blood abnormality affecting millions worldwide, requiring accurate and timely detection for effective management. This study explores the utilization of advanced DL approaches to enhance the detection of SCA in microscopic images, aiming to improve diagnostic precision and efficiency. The research involves preprocessing image data and training seven cutting-edge models: DenseNet-201, ResNet-152, Xception, MobileNet, DenseNet-201+ResNet-152, Xception+ MobileNetV2, and a comprehensive model combining DenseNet-201, ResNet-152, Xception, and MobileNet. These models are evaluated using key performance metrics, including accuracy, error rate, Jaccard score, MCC, F1-score, Kappa, and G-Mean. The study identifies the most effective model for practical application, emphasizing its potential for real-world clinical use. By providing an automated approach to early SCA detection, this research seeks to advance haematological diagnostics, improve patient care and streamline healthcare processes.