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

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Oncology
• Radiomics

Background:

• Lung cancer is a leading cause of global cancer mortality.
• Accurate segmentation of lung tissues and tumors in PET-CT images is crucial for diagnosis and treatment.
• Current methods may require significant manual effort and expertise.

Purpose of the Study:

• To investigate the efficacy of the Attention U-Net (Att-U-Net) model for segmenting lung tissues and tumors.
• To evaluate the performance of Att-U-Net on PET-CT imaging data.
• To assess the potential of AI in improving lung cancer diagnosis and treatment planning.

Main Methods:

• Utilized the Attention U-Net (Att-U-Net) deep learning model.
• Trained and evaluated the model on the Lung-PET-CT-Dx dataset.
• Measured performance using Dice Similarity Coefficient (DSC) and Intersection over Union (IoU) metrics.
• Employed Binary Cross-Entropy and Dice loss functions during training.

Main Results:

• The Att-U-Net model achieved a Dice Similarity Coefficient (DSC) of 0.81 for tumor segmentation.
• The model obtained an Intersection over Union (IoU) of 0.69 for tumor segmentation.
• Results indicate strong alignment between predicted and actual tumor regions in PET-CT images.

Conclusions:

• The Att-U-Net model demonstrates significant potential for accurate lung tumor and tissue segmentation on PET-CT scans.
• Integration into clinical workflows can enhance diagnostic accuracy and treatment planning.
• This AI approach may lead to reduced manual segmentation effort and more personalized cancer therapies.