Enhancing Lung Segmentation Algorithms to Ensure Inclusion of Juxtapleural Nodules

Luoting Zhuang, Seyed Mohammad Hossein Tabatabaei, Ashley E. Prosper, William Hsu

Abstract: Computed tomography (CT) is pivotal in detecting and monitoring pulmonary nodules in lung cancer screening. With the advancement of artificial intelligence in medical imaging, accurate lung segmentation has become crucial for reliable feature extraction. While traditional methods lack generalizability, deep learning models also encounter difficulties including juxtapleural nodules. To overcome the challenge, we finetuned a 3D U-Net by randomly masking out 70% of the images, which forces the model to infer the missing regions and learn the boundaries of the lungs. Our model achieved a Dice of 0.982 in lung segmentation. Notably, our approach achieved higher sensitivity compared to three state-of-the-art deep learning models in the inclusion of juxtapleural and large masses by 0.11, 0.20, and 0.52, respectively. Additionally, it consistently outperformed these models on external datasets. The improved result in nodule inclusion allows for more accurate and robust downstream analysis and computer-aided diagnosis of lung cancer. Our model also provided pixel-level uncertainty estimates, visually presenting where the model is confident or uncertain. High-uncertainty areas can be flagged for further examination by both clinicians and researchers. Our implementation is available at https://github.com/luotingzhuang/maskedSeg.