Robust three-dimensional registration on optical coherence tomography angiography for speckle reduction and...

Area of Science:

• Biomedical Imaging
• Medical Technology
• Ophthalmology and Dermatology

Background:

• Optical coherence tomography angiography (OCTA) uses repeated scanning and averaging to enhance image quality.
• Patient motion during scanning introduces artifacts, challenging OCTA's accuracy and reliability.
• Improved signal-to-noise ratio (SNR) is crucial for accurate visualization and quantification of vascular structures.

Purpose of the Study:

• To introduce a 3D registration method to address motion artifacts in OCT/OCTA.
• To precisely average multiple scans, improving SNR and quantification accuracy.
• To enhance the clinical utility of OCT and OCTA imaging.

Main Methods:

• A 3D registration method combining rigid affine and non-rigid B-spline transformations was developed.
• Parameter optimization utilized average stochastic gradient descent on OCT structural images.
• A multi-level resolution approach enhanced robustness and computational speed.

Main Results:

• The method demonstrated significant improvements in vessel connectivity and SNR on in vivo human skin and eye images.
• Increasing repeated volume averaging enhanced all assessed metrics: SNR, PSNR, and NCC.
• An SNR improvement from 10 to 40 dB was achieved after 10 volumetric averages.

Conclusions:

• The proposed 3D registration and averaging effectively reduces speckle noise and motion artifacts.
• The method significantly improves SNR, PSNR, and NCC metrics for averaged OCT/OCTA images.
• This algorithm is expected to enhance visualization and quantification of in vivo OCT/OCTA data for clinical applications.