Journal of Evidence-Based Medicine, Год журнала: 2025, Номер 18(1)

Опубликована: Фев. 27, 2025

ABSTRACT Objective Leukemia is a type of blood cancer that begins in the bone marrow and results high numbers abnormal white cells. Automated detection classification leukemia its subtypes using artificial intelligence (AI) machine learning (ML) algorithms plays significant role early diagnosis treatment this fatal disease. This study aimed to review synthesize research findings on AI‐based approaches from peripheral smear images. Methods A systematic literature search was conducted across four e‐databases (Web Science, PubMed, Scopus, IEEE Xplore) January 2015 March 2023 by searching keywords “Leukemia,” “Machine Learning,” “Blood Smear Image,” as well their synonyms. All original journal articles conference papers used ML detecting classifying were included. The quality assessed Qiao Quality Assessment tool. Results From 1325 identified through search, 190 studies eligible for review. mean validation accuracy (ACC) methods applied reviewed 95.38%. Among different methods, modern techniques mostly considered detect classify (60.53% studies). Supervised dominant paradigm (79% Studies utilized common methodologies classification, including preprocessing, feature extraction, selection, classification. Deep (DL) techniques, especially convolutional neural networks, most widely mentioned methodologies. Most relied internal (87%). Moreover, K‐fold cross‐validation train/test split commonly employed strategies. Conclusion are with remarkable performance. Future should prioritize rigorous external evaluate generalizability.

Язык: Английский