Fast adaptive optics scanning light ophthalmoscope retinal montaging

Area of Science:

• Ophthalmic imaging
• Optical engineering
• Computational imaging

Background:

• Adaptive optics (AO) ophthalmoscopes provide high-resolution retinal imaging.
• Image field of view is limited by the eye's isoplanatic patch, necessitating image stitching for montages.
• Current manual or automated stitching methods are time-consuming, hindering real-time analysis and large-scale studies.

Purpose of the Study:

• To develop a novel, rapid algorithm for stitching AO ophthalmoscopy images.
• To enable real-time montage generation during image acquisition.
• To improve the efficiency and applicability of AO ophthalmoscopy in research.

Main Methods:

• Image stitching using oriented fast rotated brief (ORB) descriptors and local sensitivity hashing.
• Employing a 'good enough' transformation search strategy for speed.
• Processing large datasets (250 images) for montage creation.

Main Results:

• Achieved montage processing times of 1-2 minutes per 250 images.
• Montages generated were comparable in accuracy to previous methods.
• Validated accuracy using normalized mutual information (NMI) and normalized cross-correlation (NCC).

Conclusions:

• The novel ORB-based stitching algorithm significantly reduces montage generation time.
• This rapid stitching facilitates real-time feedback and broader application of AO ophthalmoscopy.
• The method offers a practical solution for large-cohort studies and efficient ophthalmic image analysis.