Quantification and normalization of noise variance with sparsity regularization to enhance diffuse optical...

Area of Science:

• Biomedical Optics
• Medical Imaging
• Photonics

Background:

• Conventional diffuse optical tomography (DOT) reconstruction methods, often relying on Tikhonov regularization and white Gaussian noise assumptions, typically yield low spatial resolution in reconstructed images.
• This limitation hinders the diagnostic accuracy and clinical utility of DOT.

Purpose of the Study:

• To develop a novel quantification method for noise variance to improve DOT image reconstruction.
• To enhance the spatial resolution and shape fidelity of DOT images through a new algorithm.

Main Methods:

• Derived a novel quantification method for noise variance based on the linear Rytov approximation of the photon diffusion equation.
• Implemented noise variance quantification to normalize measurement signals across all source-detector channels.
• Applied sparsity regularization in conjunction with normalized signals for image reconstruction.

Main Results:

• The newly developed algorithm, quantification and normalization of noise variance with sparsity regularization (QNNVSR), significantly enhances spatial resolution and shape fidelity.
• Experimental validation using computer simulations and laboratory phantoms confirmed the effectiveness of the QNNVSR approach.
• The method demonstrated superior performance compared to conventional DOT reconstruction techniques.

Conclusions:

• The QNNVSR approach is an effective reconstruction strategy for improving DOT image quality.
• The ability to estimate noise variance with limited resources makes this method practically valuable for diverse DOT applications.
• This advancement holds promise for more accurate and reliable DOT-based diagnostics.