<p>To distinguish acute lymphoblastic leukemia (ALL) nuclei from healthy cells for early detection of leukemia, is crucial. This paper offers an Iterative Adaptive Whale Optimization Algorithm (IA-WOA) based multi-classifier system to detect impacted ALL. Firstly, the proposed IA-WOA implements Laplace distribution, Opposition-based learning (OBL) mechanism and Iterative Scaling Coefficient for enhanced investigation and exploitation. 23 State-of-the art benchmark problems of 30, 50 and 100 dimensions, respectively, are utilized to validate the performance. Results exhibit that IA-WOA achieves exact global optimal values i.e. 0 for four unimodal and two multimodal functions. Diverse analyses including Computational complexity, Wilcoxon rank-sum test, convergence and trajectories curves are carried out. The IA-WOA is compared with other metaheuristics supported by Friedman ranking statistical test. In the application part, image preprocessing is applied using Gaussian filter followed by Otsu’s inter-class variance-based image segmentation. Five categories of features—morphological, wavelet, color, texture, and statistical are extracted from the segmented nucleus images followed by LASSO method for feature selection. For classification, the IA-WOA algorithm is hybridized with standard machine learning algorithms such as Random Forest and Extra Trees to improve classification accuracy approximately from 93 to 94%, Naïve Bayes, Decision Tree, AdaBoost, Gradient Boosting Machine and K-nearest neighbors approximately from 95 to 96%, and Logistic Regression, LDA, XGBoost, and RUSBoost approximately up to 97%. Obtained results are based on confidence-interval analysis and are verified using an 80/20 hold-out test on unseen data. The data is obtained from the University of Milan's Fabio Scotti Department of Informatics website.</p>