Block matching 3D random noise filtering for absorption optical projection tomography.

Area of Science:

• Biomedical Imaging
• Developmental Biology
• Computational Imaging

Background:

• Optical projection tomography (OPT) is a 3D imaging technique valuable for developmental biology and gene expression studies.
• OPT relies on negligible light scattering in optically cleared samples, similar to X-ray CT.
• Low signal-to-noise ratio in projections hinders reconstruction quality, especially for samples with poor absorption contrast like early embryos.

Purpose of the Study:

• To improve the image quality of 3D tomographic reconstructions in optical projection tomography (OPT).
• To address challenges posed by low signal-to-noise ratio projections in OPT.
• To evaluate the effectiveness of a specific noise removal algorithm for enhancing OPT image quality.

Main Methods:

• Applied a block matching 3D (BM3D) filter to individual transillumination projection images before 3D tomographic reconstruction.
• Tested the BM3D filtering scheme on both a phantom and a real biological sample.
• Utilized standard backprojection algorithms for tomographic reconstruction after noise filtering.

Main Results:

• The BM3D filter significantly improved image quality in reconstructions with high noise levels.
• A signal-to-noise ratio increment exceeding 30 dB was achieved for severely noise-affected reconstructions.
• The filtering revealed image details previously obscured by noise in both phantom and biological samples.

Conclusions:

• The block matching 3D (BM3D) filter is effective for enhancing optical projection tomography (OPT) image quality.
• BM3D is a valuable tool for OPT, particularly under conditions of low light absorption and high noise.
• This approach offers an efficient alternative to other filtering methods like the median filter for OPT applications.