Pulmonary Nodule Detection and False Positive Reduction based on Mamba

Pulmonary nodule detection is an important step in the early diagnosis of lung cancer, but due to the diverse morphology of nodules and the complexity of CT images, false positives often occur during the detection process. This paper proposes a pulmonary nodule detection method based on the Mamba network, including a preliminary detection network and an optimized detection network. The preliminary detection network achieves efficient feature extraction and multi-scale feature fusion through deep separable convolution, dilated convolution, spatial pyramid pooling (SPP) and selective sampling module (SSM). The optimized detection network combines ensemble learning and multi-task learning strategies to further reduce false positives and improve detection accuracy. We conducted comprehensive experiments on the Luna16 dataset and verified the effectiveness of the method through the CPM evaluation indicator. The ablation experiment results show that each module has made an important contribution to improving the detection performance.