AI-Driven Innovations in Leukemia Detection: A Systematic Review of Machine Learning, Deep Learning, and Metaheuristic Techniques

Document Type : Scientific

Authors

1 Faculty of Computers and Information, Damanhur University, Egypt

2 Faculty of Science, Benha university, Egypt.

3 Faculty of Science, Benha University, Egypt

4 Faculty of Science, Benha university, Egypt

Abstract

This systematic review evaluates artificial intelligence (AI) techniques—including machine learning (ML) , deep learning (DL) , and metaheuristic optimization—in advancing leukemia detection and classification . A PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)-compliant search of Scopus, PubMed, and Web of Science (2019–2025) identified 45 high-quality studies analyzing AI applications in leukemia subtypes (e.g., acute lymphoblastic leukemia (ALL) , acute myeloid leukemia (AML) , and multiple myeloma (MM) ). Key findings reveal that DL models (e.g., convolutional neural networks (CNNs) ) achieved up to 97.2% accuracy in classifying leukemia subtypes using histopathological and flow cytometry data. Hybrid approaches like laser-induced breakdown spectroscopy (LIBS) combined with ML demonstrated 98.34% accuracy in detecting genomic markers, offering cost-effective, non-invasive solutions. Metaheuristic algorithms (e.g., binary brown-bear optimization (BBBO) ) improved feature selection, addressing high-dimensional data challenges. Notable advancements include circulating tumor DNA (ctDNA) methylation analysis (95% pre-diagnosis sensitivity) and federated learning for privacy-preserving diagnostics. However, limitations persist, such as small dataset sizes, spectral noise sensitivity in LIBS, and lack of clinical validation. Future directions include multi-center trials, integration of genomics with AI, and explainable AI to enhance clinician trust. This work highlights AI’s transformative potential in early detection and precision medicine, with implications for reducing mortality and improving patient outcomes in leukemia management.

Keywords

Main Subjects

Files

Share

How to cite

Statistics