Lung cancer segmentation Using the Att-U-Net Model on PET-CT Images

Lung cancer is the leading cause of cancer-related deaths worldwide, emphasizing the need for early and accurate diagnosis. Precise segmentation of lung regions and tumors in imaging is critical for effective diagnosis and treatment. This study investigates the Attention U-Net (Att-U-Net) model for segmenting lung tissues and tumors in positron emission tomography-computed tomography (PET-CT) images.

The Att-U-Net model was trained and evaluated for lung tissue and tumor segmentation using the publicly available Lung-PET-CT-Dx dataset from the Cancer Imaging Archive. Performance was measured using Dice Similarity Coefficient (DSC) and Intersection over Union (IoU), with Binary Cross-Entropy and Dice loss functions.

The Att-U-Net achieved a DSC of 0.81 and an IoU of 0.69 for tumor segmentation, indicating strong alignment between predicted and actual tumor regions.

The model shows high potential for accurately segmenting lung tumors and tissues on PET-CT images, supporting improved diagnosis and treatment planning.

Incorporating Att-U-Net into clinical workflows could enhance segmentation accuracy, assist clinicians in decision-making, reduce manual effort, and promote more personalized treatment strategies.