General Bayesian estimation for speckle noise reduction in optical coherence tomography retinal imagery

Area of Science:

• Biomedical Imaging
• Image Processing
• Ophthalmology

Background:

• Speckle noise in optical coherence tomography (OCT) images is a significant challenge.
• This multiplicative noise is difficult to suppress as it contains structural information.
• Effective speckle reduction is crucial for accurate analysis of OCT data.

Purpose of the Study:

• To develop a novel algorithm for speckle noise reduction in OCT images.
• To compare the performance of the proposed algorithm against existing state-of-the-art methods.
• To evaluate the algorithm's effectiveness in preserving image quality and structural details.

Main Methods:

• The algorithm projects OCT imaging data into logarithmic space.
• A Bayesian least squares estimate is derived using conditional posterior sampling.
• The method was tested on in-vivo ultrahigh-resolution OCT images of rat retinas.

Main Results:

• The proposed algorithm achieved state-of-the-art performance in speckle denoising.
• It demonstrated superior results in signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and equivalent number of looks (ENL).
• The algorithm effectively preserved edges and enhanced the visibility of fine morphological details.

Conclusions:

• The novel Bayesian-based algorithm offers effective speckle noise reduction for OCT images.
• It surpasses current methods in quantitative measures and visual quality.
• This technique improves the visualization of microstructures in retinal OCT imaging.