Automated inter-device 3D OCT image registration using deep learning and retinal layer segmentation

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Optical coherence tomography (OCT) is a key imaging technique in ophthalmology.
• Different OCT devices provide complementary data, necessitating image registration for combined analysis.
• Current registration methods may lack automation or multi-modal compatibility.

Purpose of the Study:

• To develop and evaluate a novel automated pipeline for registering multi-modal OCT images from different devices.
• To improve the accuracy and efficiency of combining complementary OCT datasets.
• To facilitate advanced clinical applications through precise OCT image registration.

Main Methods:

• A two-step automated pipeline was developed.
• Step 1: Multi-modal 2D en-face registration using deep learning.
• Step 2: Z-axis (axial) registration guided by retinal layer segmentation.

Main Results:

• The pipeline demonstrated high-quality registration performance.
• Mean error for 2D en-face registration was approximately 46 µm.
• Mean error for Z-axis registration was 9.59 µm.

Conclusions:

• The proposed automated pipeline effectively registers OCT images from different devices.
• The method achieves high accuracy in both 2D and Z-axis registration.
• This technique holds potential for improving clinical applications, including validation of retinal layer segmentation.