Constrained models for optical absorption tomography

Area of Science:

• Biomedical Optics
• Image Reconstruction
• Inverse Problems

Background:

• Optical absorption tomography faces challenges with limited data sets.
• Accurate concentration imaging is crucial for various applications.

Purpose of the Study:

• To develop an improved image reconstruction approach for optical absorption tomography using limited data.
• To enhance the accuracy and quality of concentration imaging.

Main Methods:

• Utilized near-infrared wavelength-modulated spectroscopic measurements from limited projection angles.
• Developed an image reconstruction method incorporating conventional concentration bounds and low-dimensional subspace projection.
• Reparameterized the forward model to impose spatial smoothness and inequality constraints (positivity, boundedness, logarithmic boundedness).
• Transformed the problem into an unconstrained optimization problem using surrogate parameters.

Main Results:

• The proposed method, combining affine inequality constraints and subspace projection, yielded qualitatively and quantitatively superior images compared to unconstrained reconstructions.
• Significant improvements were observed in reconstructing concentration profiles with small spatial variations.
• Gauss-Newton's algorithm demonstrated effective performance and convergence for solving the inverse problems.

Conclusions:

• The developed approach effectively addresses the inverse problem of concentration imaging in optical absorption tomography with limited data.
• The combination of inequality constraints and subspace projection offers a robust strategy for enhancing image reconstruction accuracy.
• This method shows promise for applications requiring precise concentration profiling.