Automated Blood Cell Detection and Counting

In general, medical examinations, complete blood cell (CBC) counting has been essential. Common methods, such as automated analyzers and conventional manual counting, were greatly impacted by how medical professionals operated. Deep learning algorithms for computer-aided object detection have been successfully used in various visual applications in recent years. To precisely identify and count blood cells on blood smear images, we present in this research an architecture based on deep neural networks. A publicly available BCCD (Blood Cell Count and Detection) dataset assesses our architecture's performance. Images from blood smears are frequently low resolution, with overlapping and fuzzy blood cells. Preprocessing was done on the original photos, which included blurring, sharpening, enlargement, and picture augmentation. Five models are built here with various parameters in the suggested architecture. We thoroughly examine the variables influencing their performance and evaluate how well they identify red blood cells (RBC), white blood cells (WBC), and platelets. The outcomes of the experiment demonstrate that when blood cells do not excessively overlap, our algorithms are capable of reliably identifying blood cells.