A fast nonrigid image registration with constraints on the Jacobian using large scale constrained optimization.

Area of Science:

• Medical image analysis
• Computational anatomy
• Image processing

Background:

• Accurate image registration is crucial for medical image analysis.
• Nonrigid registration is challenging due to complex deformations.
• Ensuring invertible transformations is vital for preserving image topology.

Purpose of the Study:

• To develop a novel nonrigid monomodality image registration algorithm.
• To incorporate constraints that penalize noninvertible transformations.
• To efficiently solve the optimization problem for 3D image registration.

Main Methods:

• Utilizing a cubic B-spline field to describe image deformation.
• Minimizing energy between reference and floating images.
• Employing multipliers method and L-BFGS algorithm for optimization.
• Implementing Jacobian and derivative constraints to ensure invertibility.

Main Results:

• Demonstrated effective nonrigid registration of monomodality images.
• Successfully penalized noninvertible transformations using proposed constraints.
• Efficiently handled large-scale optimization for 3D image registration.
• Validated through numerical experiments on magnetic resonance images.

Conclusions:

• The proposed B-spline based algorithm offers an effective solution for nonrigid monomodality image registration.
• The Jacobian constraints successfully prevent noninvertible transformations.
• The method is efficient and suitable for 3D medical image analysis tasks like atlas-based segmentation.