An Enhanced YOLOv11 Model for Lightweight and Efficient Precise Leukemia Detection
Jining Peng, Li Fang · 2025
The detection of medical leukemia images using deep learning techniques is a significant area of research within this field. In this paper, we introduce a novel and advantageous YOLOv11 algorithm, which belongs to the family of deep learning object detection algorithms. Tailored to the characteristics of medical images, the YOLOv11 model incorporates Depthwise Separable Convolutional Neural Networks (DWSCNN) and Residual Feature Channel Attention Modules (RFCBAM), optimizing the architecture to reduce computational load and parameter count while maintaining accuracy, thereby enhancing the model’s inference speed. Researchers have the flexibility to adjust the network’s depth and width based on specific task requirements, achieving an optimal balance between performance and efficiency. The model is trained to segment and label different stages of leukemia, categorized as benign, pre-malignant, and post-malignant. Experimental results demonstrate that the proposed model achieves a precision of $98.6 \%$ on this segmented dataset, significantly outperforming the YOLOv8 model and other leading techniques. This work not only underscores the effectiveness and precision of the proposed model in detecting leukemia images but also provides innovative perspectives for the future application of object detection in medical imaging and clinical disease diagnosis.