Formulation of image fusion as a constrained least squares optimization problem.

Area of Science:

• Medical Imaging
• Computer Vision
• Optimization

Background:

• Image fusion combines lower resolution color and higher resolution monochrome images in medical diagnostics.
• This process aims to enhance spatial context and signal-to-noise ratio for better clinical information.
• Current methods often involve complex image decomposition techniques.

Purpose of the Study:

• To formulate medical image fusion as a convex optimization problem.
• To develop an efficient and parallelizable algorithm for image fusion.
• To avoid image decomposition and enable pixel-level operations.

Main Methods:

• Image fusion is framed as a convex optimization problem.
• The method operates directly at the pixel level, avoiding decomposition.
• Leverages robust and simple numerical methods for global minimization.

Main Results:

• Achieved a highly efficient and embarrassingly parallelizable algorithm.
• The fused image is realized as the global minimizer of the convex problem.
• Demonstrates a novel approach to medical image fusion.

Conclusions:

• The proposed convex optimization framework offers an efficient solution for fusing medical images.
• Pixel-level operations and avoidance of decomposition lead to a robust algorithm.
• This method provides clinicians with a single, information-rich frame for improved diagnosis.