AI-Based Automated Classification of Acute Lymphoblastic Leukemia from Peripheral Blood Smear Images: A Comprehensive Review
Main Article Content
Abstract
Acute lymphoblastic leukemia (ALL) is one of the most frequent malignancies in children and proper identification at early stage is critical for better clinical outcomes. However, the manual review of peripheral blood smear (PBS) images is time consuming, subjective and relies upon expert hematological interpretation. This detailed review discusses AI methods developed for the identification and categorization of ALL based on PBS pictures. Recent supervised deep learning models have demonstrated high classification accuracy on benchmark datasets. However, the interpretation of these results should be done with caution, as many studies are based on limited datasets, controlled acquisition conditions, insufficient external validation, and possible dataset bias. A major drawback in this sector is the lack of expert annotated clinical images, which raises annotation cost and might decrease model generalizability across laboratories, staining processes and patient groups. Methods of semi-supervised learning (SSL) such as FixMatch and consistency regularization provide a promising route by integrating a small batch of labeled data with a larger pool of unlabeled images. The proposed FixMatch-EfficientNet-B3 framework is simply a conceptual future research direction, not an experimentally verified diagnostic model. This review points out the necessity for multi-center datasets, clear evaluation methodologies, computational feasibility analysis and reliable validation before the translation of AI-based ALL classification systems into practical diagnostic practice.
Downloads
Article Details
Issue
Section

This work is licensed under a Creative Commons Attribution 4.0 International License.