A New Smartphone-Based Optic Nerve Head Biometric for Verification and Change Detection

Area of Science:

• Ophthalmology
• Biometrics
• Artificial Intelligence

Background:

• Smartphones with lens adaptors are replacing traditional ophthalmoscopes.
• Glaucoma and diabetic retinopathy cause preventable blindness, often with asymptomatic optic nerve head (ONH) changes.
• Developing regions face ophthalmologist shortages but have widespread mobile phone access.

Purpose of the Study:

• To develop a proof-of-concept optic nerve head (ONH) biometric application for smartphones.
• To utilize unique ONH blood vessel patterns for routine biometric verification.
• To enable early detection of asymptomatic ONH changes indicative of serious eye diseases.

Main Methods:

• Developed the iKey application platform using three deep neural networks (DNNs).
• Trained blood vessel specific feature (BVSF) and graticule blood vessel (GBV) DNNs on unique blood vessel vectors.
• Used a non-feature specific (NFS) ResNet50 DNN as a baseline for comparison.

Main Results:

• The BVSF DNN achieved a verification accuracy of 97.06%.
• The GBV DNN reached 87.24% accuracy.
• The NFS DNN provided a baseline accuracy of 79.8%.

Conclusions:

• A novel ONH biometric system was created using a hybrid platform of ONH algorithms for smartphone verification.
• The system alerts users to potential image changes, facilitating early observation of silent disease progression.
• The application stores ONH image history for future longitudinal analysis of biometric changes.