A Hybrid Deep Learning Framework for Accurate Blood Cell Counting and Classification Using Advanced Segmentation and Feature Extraction

Abstract

Blood cell counting is a crucial process in medical diagnostics for identifying and classifying red blood cells (RBCs), white blood cells (WBCs), and platelets. Accurate counting and classification are challenging due to overlapping cells, noise, and variations in cell shapes. To overcome these issues, we propose a robust approach integrating Segmentation using Canny Edge Detector and Watershed Segmentation. For Feature Extraction, we employ Texture-Based Features using Local Binary Patterns (LBP). Finally, Counting and Classification are performed using Convolutional Neural Networks (CNN). This combined method aims to enhance accuracy and efficiency in automated blood cell analysis.