Machine Learning in Detection and Classification of Leukemia Using Smear Blood Images: A Systematic Review

Introduction. The early detection and diagnosis of leukemia, i.e., the precise differentiation malignant leukocytes with minimum costs in stages disease, is a major problem domain disease diagnosis. Despite high prevalence there shortage flow cytometry equipment, methods available at laboratory diagnostic centers are time-consuming. Motivated by capabilities machine learning (machine (ML)) diagnosis, present systematic review was conducted to studies aiming discover classify leukemia using learning. Methods. A search four databases (PubMed, Scopus, Web Science, ScienceDirect) Google Scholar performed via strategy Machine Learning (ML), peripheral blood smear (PBS) image, detection, classification as keywords. Initially, 116 articles were retrieved. After applying inclusion exclusion criteria, 16 remained population study. Results. This study presents comprehensive view status all published ML-based models that process PBS images. average accuracy ML applied image analysis detect >97%, indicating use could lead extraordinary outcomes from Among techniques, deep (DL) achieved higher precision sensitivity detecting different cases compared its precedents. has many applications analyzing types images, but algorithms acute lymphoblastic (ALL) attracted greatest attention fields hematology artificial intelligence. Conclusion. Using method images can improve accuracy, reduce time, provide faster, cheaper, safer services. In addition current methods, clinical experts also adopt tools.