MCUNet:Multi-Level Feature and Uncertainty-Guided Lung Nodule Segmentation

Lung cancer ranks among the top causes of cancer-related mortality worldwide, and precise segmentation of lung nodules is essential for early detection and diagnosis. However, the low contrast of small nodules often makes them difficult to distinguish from noise and other structures in medical images, posing a significant challenge for accurate segmentation. This paper presents a novel lung nodule segmentation network, the Multi-Level Feature and Uncertainty-Guided Lung Nodule Segmentation Network (MCU-Net). MCU-Net includes a Multi-Level Dense Feature Extraction Module, an Uncertainty-Aware Module, and a Feature-Aware Attention Module to enhance the accuracy and robustness of lung nodule segmentation. By combining multi-scale feature extraction and uncertainty-aware learning, the proposed network achieves superior performance on the LIDC-IDRI dataset, with an average Dice score of 86.16