Merging parametric active contours within homogeneous image regions for MRI-based lung segmentation.

Area of Science:

• Medical Imaging
• Biophysics
• Computational Anatomy

Background:

• Hyperpolarized helium-3 (3He) magnetic resonance (MR) imaging is used to assess lung functionality.
• Accurate segmentation of lung cavities from traditional proton (1H) MR images is crucial for analyzing 3He MR data.
• Existing segmentation methods struggle with the complex structures of lung cavities.

Purpose of the Study:

• To develop an advanced active contour model for precise lung cavity segmentation from 1H MR images.
• To enable accurate evaluation of total lung ventilation using hyperpolarized 3He MR imaging.
• To introduce a novel method for merging multiple contours in image segmentation.

Main Methods:

• Development of a parametric active contour model for lung cavity segmentation.
• Implementation of a merging technique utilizing an external force field derived from partial differential equations.
• Establishing a theoretical link between the force field and fluid flow in porous media.

Main Results:

• The active contour model accurately captures high-curvature features and provides smooth boundaries for lung cavities.
• The proposed merging method effectively combines multiple contours, stopping accurately at object boundaries.
• Experimental validation on synthetic and lung 1H MR images demonstrated successful segmentation.

Conclusions:

• The developed active contour model offers a robust solution for segmenting lung air spaces in 1H MR images.
• This segmentation technique is vital for computing functional air space from hyperpolarized 3He MR images.
• The method advances the analysis of lung function through improved MR image processing.