Retinal Vascular Biometrics for Personal Identification in Forensic Investigations: A Pilot Study in the Palembang Population

Abstract

Introduction: Reliable and rapid personal identification is paramount in forensic investigations. Traditional methods can be time-consuming or challenging, particularly in cases involving fragmented remains or limited access to comparative data. Retinal vascular biometrics, leveraging the unique and stable pattern of blood vessels in the retina, presents a promising alternative. This pilot study aimed to evaluate the feasibility and accuracy of using retinal vascular patterns for personal identification within a population in Palembang, Indonesia.

Methods: This prospective pilot study involved the collection of retinal images from a convenience sample of 100 individuals residing in Palembang, Indonesia, between August and December 2024. Retinal images were captured using a non-mydriatic fundus camera model commonly available in ophthalmological clinics. Pre-processing steps included image enhancement and noise reduction. Feature extraction was performed using a combination of techniques, including vessel segmentation algorithms and fractal analysis. A matching algorithm based on normalized cross-correlation and feature vector distance was employed to compare retinal images. The performance of the biometric system was evaluated using metrics such as the False Acceptance Rate (FAR), False Rejection Rate (FRR), and Equal Error Rate (EER).

Results: The analysis of 10000 comparison attempts (100 genuine and 9900 imposters) yielded promising results. The calculated EER for the retinal vascular biometric system was 0.85%. The FAR at a 0% FRR was 0.1%, and the FRR at a 0% FAR was 1.6%. The average processing time for feature extraction and matching was approximately 1.5 seconds per comparison. Demographic analysis suggested no significant difference in accuracy across different age groups within the studied sample.

Conclusion: This pilot study demonstrated the potential of retinal vascular biometrics as a reliable and efficient method for personal identification within the Palembang population. The low EER suggests a high level of accuracy. Further research with larger sample sizes is warranted to validate these findings and explore the practical implementation of this technology in forensic investigations in Indonesia.