X-Let's Atom Combinations for Modeling and Denoising of OCT Images by Modified Morphological Component Analysis

Area of Science:

• Medical Imaging
• Image Processing
• Ophthalmology

Background:

• Optical Coherence Tomography (OCT) is crucial for diagnosing retinal abnormalities.
• Speckle noise significantly degrades OCT image quality, hindering accurate analysis.
• Existing denoising methods often fall short in preserving OCT image details.

Purpose of the Study:

• To develop an improved method for denoising OCT retinal images.
• To enhance the diagnostic utility of OCT by reducing speckle noise.
• To provide interpretable image components for further processing tasks like classification.

Main Methods:

• Employed Modified Morphological Component Analysis (MMCA) for image decomposition.
• Utilized non-data-adaptive multi-scale (X-let) transforms to create suitable dictionaries.
• Applied adaptive local thresholding to individual image components for denoising.

Main Results:

• Achieved significantly improved denoising performance compared to state-of-the-art algorithms.
• Demonstrated superior results in both Peak Signal-to-Noise Ratio (PSNR) and no-reference image quality assessments.
• Successfully decomposed OCT images into interpretable components.

Conclusions:

• The proposed MMCA method offers effective speckle noise reduction in OCT images.
• The denoising enhances image quality for better retinal abnormality diagnosis.
• The image decomposition facilitates subsequent image analysis and classification tasks.