Abstract
The segmentation of lung nodules is a crucial step in the early detection of lung cancer, which remains the leading cause of cancer-related mortality worldwide. To address the need for efficient and accurate diagnosis, we introduce CSEA-Net, a fully automated lung nodule segmentation model designed to operate with high precision on computed tomography images. The model employs a deep learning architecture that incorporates the dual-branch channel-spatial feature enhancement network and a coordinate attention mechanism to address the diagnostic challenges posed by lung nodules that are too small and poorly contoured. CSEA-Net achieves excellent performance in lung tumor segmentation across multiple publicly available datasets with high Dice coefficients and Intersection over Union scores. In conclusion, our proposed model exhibits excellent performance in the accurate segmentation of lung nodules and has the potential to assist doctors in automatically distinguishing the locations of lung tumors.