Traditional pathology utilizes stained blood smears viewed through microscopy, a technique still prevalent despite technological advancements. Typically, manual human assessment accompanies automated methods, maintaining simplicity but lacking comprehensive cell evaluation. In an effort to enhance this approach, we employed AI-driven automation through the web platform IKOSA. Our methodology, applied to DAPI-Giemsa co-stained blood smears, successfully segmented and identified various blood cell types, including neutrophils, lymphocytes, eosinophils, monocytes, erythrocytes, and platelets. Unlike previous algorithms focusing on singular cell types, our system provides quantitative measurements and sophisticated analyses, utilizing entropy and gray-level co-occurrence matrices. This innovative approach surpasses classical methods, offering potential for monitoring cellular structural changes associated with diseases or treatment responses. In conclusion, AI-based automation in blood cell evaluation presents an opportunity to enhance routine diagnostics by incorporating quantitative parameters into traditional leukocyte counts.
View the full article in the journal and see how IKOSA supported the research