[Research progress in lung parenchyma segmentation based on computed tomography].

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Pulmonology

Background:

• Lung diseases like cancer and COVID-19 pose significant health risks, necessitating early detection.
• Computed tomography (CT) is vital for lung disease screening, with lung parenchyma segmentation being a critical step.
• Manual segmentation is time-consuming and subjective, highlighting the need for automated methods.

Purpose of the Study:

• To review recent advancements in lung parenchyma segmentation techniques using CT images.
• To compare traditional machine learning and deep learning approaches for lung segmentation.
• To identify current challenges and future prospects in automated lung parenchyma segmentation.

Main Methods:

• Literature review of domestic and international research on lung parenchyma segmentation.
• Comparative analysis of traditional machine learning and deep learning methodologies.
• Emphasis on improvements in deep learning network architectures for enhanced segmentation.

Main Results:

• Deep learning methods show significant promise for automatic, fast, and accurate lung parenchyma segmentation.
• Advancements in network structures have improved segmentation quality compared to traditional methods.
• The review synthesizes current research, offering insights into method efficacy.

Conclusions:

• Automated lung parenchyma segmentation is a key research area for improving early lung disease diagnosis.
• Deep learning models offer superior performance over traditional methods for segmentation tasks.
• Further research into network optimization and addressing segmentation challenges is warranted.