Nonrigid medical image registration based on mesh deformation constraints

Area of Science:

• Medical Imaging
• Computational Anatomy
• Biomedical Engineering

Background:

• Nonrigid medical image registration is crucial for comparing anatomical structures.
• Regularizing the deformation field is essential for accurate and reliable registration.
• Existing methods may face challenges in preserving topological consistency.

Purpose of the Study:

• To propose a novel regularization constraint for nonrigid medical image registration.
• To introduce a method utilizing a triangular mesh and spring analogy for deformation field regularization.
• To evaluate the proposed method's performance in registering cerebral magnetic resonance imaging (MRI) data.

Main Methods:

• Covering the template image with a triangular mesh.
• Applying a regularization constraint based on connections between mesh vertices.
• Utilizing the spring analogy to preserve the connection relationship within the mesh.
• Evaluating the method on inter-individual cerebral MRI data.

Main Results:

• The proposed method demonstrates good deformation ability.
• The method exhibits strong topology-preserving capabilities.
• Successful registration of cerebral MRI data from different individuals was achieved.

Conclusions:

• The developed mesh-based regularization offers a new approach for nonrigid medical image registration.
• The spring analogy effectively preserves mesh connectivity, enhancing registration accuracy.
• The method shows promise for applications requiring robust and topologically consistent image registration.