Improving Medical Diagnostics with EfficientNetV2-B1 in Blood Cell Analysis

This research paper investigates how deep learning can be applied to classify blood cells, focusing on the architecture of EffectiveNetV2B1. It took a total of 17,092 images to make up the dataset for this study, which were arranged by blood cell, eight groups by and into each. Angle types were white cells Neutrophils, Eosinophils, Basophils Lymphocytes Monocytes Immature Granulocytes, (including promyelocytes for the latter three and myelocytes metamyelocytes), Erythroblasts and platelets or thrombocytes. The mainstay of our method is the training process, and over 40 epochs EfficientNetV2B1 training let us to display its strength. After screening Examination in a comprehensive range of tests, our model shows exemplary performance, achieving an accuracy rate of 92%. This can be further shown by the remarkable accuracy of its Receiver Operating Characteristic (ROC) value: 99.46%. These findings underscore the effectiveness of deep learning models, especially the EfficientNetV2B1 architecture, in medical image analysis and classification. Our research makes an important contribution to diagnostics in medicine by identifying blood cell types with precision and contains information for medicine, potentially helping care providers diagnose a variety of blood-associated diseases in more efficient ways than before.