Pulmonary nodule detection model based on YOLOv8-MSDA

To solve the problems of missed diagnoses and misdiagnoses caused by traditional methods' reliance on manual interpretation in pulmonary nodule detection, an improved YOLOv8 model integrating deep learning is provided in this paper. By implementing a cross-layer feature pyramid network for multi-scale feature fusion, integrating an MSDA attention mechanism to enhance the perception of pulmonary nodules, and adopting CLAME image enhancement technology to improve contrast, the model significantly improves detection accuracy and efficiency. Experimental results demonstrate that this method can effectively identify pulmonary nodules in lung CT images, achieving a recall rate of 0.93 and mAP of 89.7% on the LIDC-IDRI data set. This detection method can effectively identify small pulmonary nodules and those in complex backgrounds, providing a reliable tool for early lung cancer screening. There are significant implications for optimizing the allocation of healthcare resources.