A Parallel Fusion of CNN and Transformer for CT Lung Nodule Detection

To improve the precision of CT lung nodule detection, this paper presents a parallel fusion model based on CNN and Transformer network, which integrates features of the two networks to fully utilize the high-resolution spatial information of CNN and the global semantic information of Transformer, thereby enhancing the model's expressive capability and providing high-quality input for downstream tasks. Experimental results show that the proposed improves the sensitivity of detection and reduce the false positive rate based on anatomical features, while also considering algorithm efficiency.