Automatic left and right lung separation using free-formed surface fitting on volumetric CT.

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Thoracic Imaging

Background:

• Accurate lung segmentation is crucial for quantitative analysis in thoracic imaging.
• Manual segmentation of left and right lungs from volumetric CT data is time-consuming and subject to inter-observer variability.

Purpose of the Study:

• To develop and validate a fully automated method for separating left and right lung volumes using free-formed surface fitting on computed tomography (CT) images.
• To assess the performance of the automated method against expert manual segmentation.

Main Methods:

• Iterative 3D morphological operations and Hessian matrix analysis for initial lung division.
• Euclidean distance transform to detect an optimal separating surface between lung regions.
• Free-formed surface fitting algorithm to model the separating surface for precise lung separation.

Main Results:

• The automated method achieved high accuracy, with average root mean square surface distance of 0.032 mm and maximum surface distance of 2.418 mm.
• Volumetric overlap error between automated and manual segmentation was minimal at 0.017%.
• The method demonstrated feasibility in accurately separating left and right lung volumes on volumetric chest CT.

Conclusions:

• The proposed automated free-formed surface fitting method provides a robust and accurate solution for left and right lung separation in CT imaging.
• This technique has the potential to streamline quantitative analysis in thoracic imaging and improve diagnostic efficiency.
• The high agreement with manual segmentation suggests clinical applicability for automated lung volume assessment.