NerveStitcher: Corneal confocal microscope images stitching with neural networks

Area of Science:

• Ophthalmology
• Medical Imaging
• Computational Neuroscience

Background:

• Corneal nerves are crucial biomarkers for early neurological disorder diagnosis.
• In vivo confocal microscopy (IVCM) is a key tool for observing and quantifying corneal nerves.
• Stitching multiple IVCM images into a wide-field montage aids in comprehensive corneal nerve plexus analysis.

Purpose of the Study:

• To develop a fully automatic image stitching method for IVCM images using neural networks.
• To improve the accuracy and efficiency of corneal nerve morphology measurement.
• To enhance the reliability of patient corneal nerve evaluation.

Main Methods:

• Utilized a self-supervised point detector to identify feature points in IVCM images.
• Implemented an attention-based mechanism for flexible point correspondence and partial image pair assignment.
• Integrated and fused scattered IVCM images based on local offset calculations.

Main Results:

• The proposed neural network-based stitching method demonstrated improved matching accuracy compared to conventional techniques.
• Significantly reduced processing time for image stitching.
• Enabled more accurate and reliable calculation of corneal nerve morphological parameters.

Conclusions:

• The automatic stitching method provides a more precise and efficient approach for analyzing corneal nerve morphology from IVCM images.
• This advancement supports more accurate diagnosis and monitoring of neurological conditions through corneal nerve assessment.
• The developed method offers a valuable tool for both clinical and research applications in ophthalmology and neuroscience.