DRIXen Architecture: An Ensemble-Driven Classification of Acute Lymphoblastic Leukemia

Acute Lymphoblastic Leukemia (ALL) is an aggressive blood cancer affecting immature lymphocytes, requiring timely and accurate diagnosis. Traditional methods are slow, expertise-dependent, and prone to errors. We propose DRIXen, a custom deep learning model that generates specialized variants from DenseNet-169, ResNet-101, Inception-V3, and Xception, which are further ensembled into two final models, EN-1 and EN-2, to enhance overall performance across multiple evaluation metrics. Using C-NMC 2019 and ALL-IDB2 datasets with 5-fold cross-validation, our approach ensures robustness and generalization. Our model achieves state-of-the-art performance, with 99.85% (EN-1) and 99.93% (EN-2) on C-NMC 2019 and 99.86% (EN-1) and 99.87% (EN-2) on ALL-IDB2, surpassing existing methods. These findings highlight the potential of our approach to improve leukemia diagnosis by offering an accurate, efficient, and automated classification system for clinical applications.