Patch-based anisotropic diffusion scheme for fluorescence diffuse optical tomography--part 2: image reconstruction.

Area of Science:

• Biomedical imaging
• Optical tomography
• Image reconstruction

Background:

• Fluorescence diffuse optical tomography (fDOT) generates 3D images of molecular and cellular processes in biological tissues.
• Image reconstruction in fDOT is ill-posed, necessitating regularisation techniques for stable and meaningful solutions.
• Existing quadratic regularisation methods often result in oversmoothing or noisy images, while anisotropic diffusion (AD) struggles to differentiate edges from noise.

Purpose of the Study:

• To develop and evaluate a novel regularisation method for fDOT image reconstruction.
• To improve image quality, preserve features, and reduce noise in fDOT images.
• To enhance the resolution and accuracy of fluorescence distribution imaging.

Main Methods:

• Proposed patch-based anisotropic diffusion regularisation (PAD) using patch similarity for weighting.
• Introduced wavelet compression (PAD-WT) to reduce computational complexity and leverage denoising properties.
• Incorporated structural information from anatomical images into PAD-WT for enhanced feature preservation and resolution.

Main Results:

• PAD-WT demonstrated superior performance in denoising simulated and in vivo mouse fDOT data.
• The method produced high-quality, noise-free images, outperforming standard AD techniques.
• Incorporating structural information improved image quality and resolution, preserving fluorescence signals not visible in anatomical images.

Conclusions:

• PAD-WT is an effective regularisation technique for fDOT image reconstruction.
• The method offers significant improvements in image quality, noise reduction, and feature preservation.
• PAD-WT holds promise for advancing molecular and cellular imaging applications using fDOT.