Classification of Human Peripheral Blood Cells Using EfficientNet B0-B7 Models

Peripheral blood cells are essential for the human immune system's defense against microorganisms that cause disease. Precise classification of these cells is essential for medical diagnostics since they contain significant information that may be evaluated to assess an individual's health status. Accurate blood cell type classification is critical to histopathology because it offers valuable information for detecting diseases like viruses and cancers. In this work, we propose an advanced deep learning (DL) framework based on the EfficientNet series (B0 to B7) for the automatic categorization of peripheral blood cells. Our approach targets eight distinct classes, including basophils, eosinophils, neutrophils, lymphocytes, monocytes, immature granules (IG), erythroblasts and platelets, utilizing two comprehensive datasets: the publicly available Raabin Health Database and the PBC dataset. The proposed EfficientNet models demonstrate superior performance, achieving highly accurate classification of blood cells with significantly fewer epochs and reduced computational time compared to alternative methods. Our method's performance is thoroughly assessed, demonstrating its efficiency as a reliable tool for blood cell analysis in medical applications.