Publisher

STIKESBUDILUHURCIMAHISTIKESBUDILUHURCIMAHI

Journal JKBLJKBL Abstract

White blood cell (WBC) classification plays a crucial role in hematological diagnosis and is typically performed manually using microscopic images. However, manual analysis is limited by subjectivity and time inefficiency. With recent technological advances, artificial intelligence (AI) offers promising solutions for automated WBC classification that enhance accuracy and efficiency. This study presents a scoping review of 20 scientific publications discussing AI applications in microscopic image-based WBC classification. Literature searches were conducted in PubMed, ScienceDirect, Institute of Electrical and Electronics Engineers (IEEE) Xplore, and Google Scholar using relevant keywords such as “AI, “white blood cell, and “microscopic image. Findings indicate that the most commonly used method is Convolutional Neural Network (CNN), either standalone or hybrid (e.g., YOLOv5, ResNet, Vision Transformer), achieving accuracies up to 99.7%. The datasets were mostly public Blood Cell Count and Detection (BCCD), Leucocyte Images for Segmentation and Classification (LISC), Raabin-WBC or local laboratory sources. The reviewed studies aimed at automatic WBC detection, classification, and morphological identification. Despite encouraging outcomes, challenges such as external validation and limited access to real clinical data remain. Overall, AI has proven effective in enhancing speed, accuracy, and objectivity in WBC classification. Further research is needed to support AI integration into real-world clinical laboratory practice.

Conclusion

This scoping review demonstrates the effectiveness of deep learning models, particularly CNNs, in enhancing the accuracy and efficiency of white blood cell classification using microscopic images.While AI shows promise in improving objectivity and reproducibility in hematological diagnostics, challenges related to external validation and the limited availability of diverse clinical data remain.Ultimately, the successful integration of AI into clinical practice requires strategic efforts focused on generalizability, interpretability, and adherence to regulatory standards.

Future Research

Future research should prioritize the creation and standardization of larger, more diverse datasets of microscopic blood cell images, encompassing various populations to improve the generalizability of AI models. Furthermore, rigorous external validation studies are crucial to assess the performance of these models across different laboratory settings and patient demographics, ensuring their reliability in real-world clinical applications. Finally, developing user-friendly interfaces and comprehensive training programs for healthcare professionals will be essential to facilitate the seamless integration of AI-powered tools into routine hematological workflows, ultimately enhancing diagnostic accuracy and efficiency while minimizing the potential for misinterpretation and maximizing clinical impact. These advancements will pave the way for more objective, efficient, and accessible hematological diagnostics, benefiting both clinicians and patients.

References

1. Klasifikasi Sel Darah Putih Menggunakan Metode Support Vector Machine (SVM) Berbasis Pengolahan Citra... journal.ugm.ac.id/ijeis/article/view/15420Klasifikasi Sel Darah Putih Menggunakan Metode Support Vector Machine SVM Berbasis Pengolahan Citra journal ugm ac ijeis article view 15420

Metadata

File size	196.88 KB
Pages	8
DMCA	Report