Quantitative 3D-OCT motion correction with tilt and illumination correction, robust similarity measure and...

Area of Science:

• Ophthalmology
• Medical Imaging
• Biomedical Engineering

Background:

• 3D Optical Coherence Tomography (OCT) motion correction algorithms face challenges from variable illumination, signal quality, tilt, and motion.
• Existing post-processing methods struggle with these artifacts, impacting the accuracy of quantitative measurements.

Purpose of the Study:

• To develop and evaluate an advanced 3D-OCT motion correction algorithm.
• To address challenges in illumination variability, signal quality, tilt, and motion for improved 3D-OCT data analysis.

Main Methods:

• An advanced 3D-OCT motion correction algorithm utilizing image registration and orthogonal raster scan patterns.
• Incorporation of a pseudo Huber norm for intensity similarity and a pseudo L0.5 norm for regularization.
• A two-stage registration process: coarse correction of axial motion/tilt followed by full optimization.

Main Results:

• The advanced algorithm significantly improved reproducibility compared to a basic algorithm and no correction.
• Mean absolute retinal thickness difference reduced from 9.9 um (no correction) to 5.0 um (advanced algorithm).
• Blood vessel likelihood map error reduced to 47% of the uncorrected error using the advanced algorithm.

Conclusions:

• The developed advanced motion correction algorithm substantially enhances the reliability of quantitative measurements from 3D-OCT data.
• The algorithm demonstrates superior performance over basic methods, offering potential for more accurate ophthalmic diagnostics.