SpringLS: a deformable model representation to provide interoperability between meshes and level sets.

Area of Science:

• Computer vision
• Medical imaging
• Computational geometry

Background:

• Traditional deformable models often use either mesh or level set representations, limiting interoperability between different computational methods.
• Integrating these representations is challenging due to inherent differences in their mathematical formulations and data structures.

Purpose of the Study:

• To introduce a novel deformable model, the Spring Level Set (SpringLS), that merges mesh and level set representations into a unified framework.
• To demonstrate the interoperability and shape information preservation capabilities of the SpringLS model.

Main Methods:

• The Spring Level Set (SpringLS) model utilizes a constellation of triangular surface elements, termed 'springls', to define a level set.
• The SpringLS representation can be interpreted and utilized as either a mesh or a level set, ensuring no loss of shape information during transformation.

Main Results:

• The SpringLS model successfully integrated mesh and level set functionalities, offering a unified deformable model.
• Demonstrated successful application in joint segmentation and spherical mapping of the human brain cortex.
• Showcased effectiveness in atlas-based and non-atlas-based segmentation of pelvic structures.

Conclusions:

• The Spring Level Set (SpringLS) provides a versatile and unified deformable modeling approach.
• This unified representation enhances interoperability between mesh-based and level set-based methods in medical image analysis and other fields.
• The model preserves shape information, enabling accurate segmentation and mapping tasks.