Advanced YOLO V11: Integrating Multi-Level Feature Fusion and Attention Mechanisms for Enhanced Blood Cell Detection Precision

Automatic blood cell detection is an essential part of hematological diagnosis. However, achieving high accuracy in fully automatic detection remains challenging due to microscopic image complexity and cell size variations. This paper proposes a new detection framework that endows YOLO with hierarchical feature fusion and attention mechanism capabilities. Experimental verification on the standardized Complete Blood Cell Count dataset demonstrated a 1.5% improvement in the detection precision brought by the hierarchical feature fusion strategy, whereas the attention mechanism contributed an extra 2.3% increase in precision metrics. Although the overall performance of the proposed framework for white blood cell detection was superior, accuracy varied for different cell categories. Though the computational requirements increased, the increase in detection capability outweighs the limitation. These findings advance the understanding of feature integration in medical image analysis, with practical implications for improvement in clinical diagnostic efficiency. The methodology laid down in this study offers promising directions for the advancement of automated hematological analytical systems.