Hybrid Ensemble Methods for Acute Lymphoblastic Leukemia Diagnosis: Integrating CNN Features with Artificial Rabbits Optimization Algorithm

Acute Lymphoblastic Leukemia (ALL) is a rapidly advancing blood cancer that necessitates timely and accurate diagnosis for effective treatment. This project presents a hybrid approach for ALL classification by utilizing advanced methods for feature extraction and integration. The DenseNet121, ResNet101 and MobileNet models are employed to extract various features from microscopic blood smear images. Principal Component Analysis (PCA) is used to select the most relevant features, which are then combined using advanced integration techniques to merge complementary information. To further improve the performance of the integrated features, the Artificial Rabbits Optimization (ARO) algorithm is applied to refine the feature set by selecting the most discriminative attributes. The optimized feature set is then passed through Random Forest (RF) and XGBoost classifiers, resulting in precise and reliable predictions. This hybrid framework achieves a classification accuracy of 0.96, demonstrating its potential for accurate detection and differentiation of ALL cells from healthy cells.