Interretinal Symmetry in Color Fundus Photographs

Area of Science:

• Ophthalmology
• Biometrics
• Computer Vision

Background:

• Bilateral symmetry is evident externally in human eyes, but internal retinal symmetry is less understood.
• The retina, crucial for vision, also presents potential as a biometric identifier.
• Previous research has not extensively explored interretinal symmetry from a biometric verification standpoint.

Purpose of the Study:

• To determine if approximate bilateral symmetry exists within the retina.
• To assess the viability of using interretinal symmetry for biometric identification.
• To evaluate the effectiveness of human and neural network-based methods for retinal verification.

Main Methods:

• Analysis of overall retinal symmetry, not specific features like vessel count or thickness.
• Evaluation of human performance in bilateral retina verification using fundus images.
• Assessment of neural network performance in bilateral retina verification using fundus images.
• Experimentation on a publicly available dataset.

Main Results:

• Interretinal symmetry was confirmed through experimentation.
• Both human and neural network approaches demonstrated capabilities in bilateral retina verification.
• The study provides evidence for the biometric potential of retinal symmetry.

Conclusions:

• Approximate bilateral symmetry is present in the human retina.
• Interretinal symmetry can be leveraged for reliable biometric identification.
• Further research into retinal symmetry can enhance biometric security systems.