Pleura detection in chest computed tomography with application for nodule detection

Area of Science:

• Medical Imaging
• Computational Anatomy
• Pulmonary Medicine

Background:

• Accurate segmentation of pulmonary nodules is crucial for diagnosis and treatment planning.
• Identifying nodules in contact with pleural surfaces presents a segmentation challenge due to complex anatomical interfaces.

Purpose of the Study:

• To develop and validate an efficient algorithm for detecting pleural objects adjacent to pulmonary nodules.
• To automatically separate nodules from pleural surfaces (chest wall, diaphragm) when contact occurs.

Main Methods:

• A recursive curve-fitting algorithm was applied to each slice of the region of interest.
• The algorithm located objects between parietal and visceral pleurae and assessed curve-fitting quality.
• Nodule separation from pleural surfaces was automated using the fitting curve.

Main Results:

• The algorithm was tested on 864 slices from 40 nodules.
• Visual inspection by attending physicians reviewed segmentation accuracy.
• A consensus accepted 93.6% of the segmentation results, indicating high performance.

Conclusions:

• The proposed algorithm efficiently detects pleural objects in contact with nodules.
• The automated separation of nodules from pleural surfaces is effective and accurate.
• This method shows significant potential for improving pulmonary nodule analysis in medical imaging.