B2E-CDG: Conditional diffusion-based for label-free OCT angiography artifact removal and robust vascular...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Ophthalmology

Background:

• Optical Coherence Tomography Angiography (OCTA) is crucial for diagnosing various diseases.
• Motion artifacts from patient movement degrade OCTA image quality and diagnostic accuracy.
• Existing artifact removal methods require extensive labeled data and complex dataset generation.

Purpose of the Study:

• To develop a novel method for removing motion artifacts from OCTA images.
• To leverage OCTA B-scan structural and flow information often under-utilized in enface images.
• To eliminate the need for labeled datasets and pseudo-stripe generation in artifact correction.

Main Methods:

• Proposed B-scans to Enface Conditional Diffusion Guidance (B2E-CDG) to translate signal-void B-scans into normal B-scans.
• Utilized conditional guidance within a diffusion model, incorporating normal and reference B-scans for style feature guidance.
• Exploited inherent paired signal-void and normal B-scans from OCTA's repetitive scanning nature.

Main Results:

• B2E-CDG effectively removed motion artifacts from OCTA images.
• The method successfully restored critical vascular and structural details.
• Demonstrated superior performance in vascular recovery and artifact removal metrics compared to existing methods.

Conclusions:

• B2E-CDG offers an efficient and effective solution for OCTA motion artifact correction.
• The proposed method enhances the clinical utility and diagnostic reliability of OCTA imaging.
• This approach avoids the limitations of labeled datasets and pseudo-stripe generation inherent in other methods.