Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study introduces a retinal biometric system designed to verify patient identities and retrieve accurate records from color fundus images, addressing critical identity errors that can affect medical records and clinical decisions. Utilizing a 512-dimensional metric-learning encoder with a ConvNeXtV2 backbone, the system was trained on a substantial dataset of over 227,000 images from diverse patient-eye identities, demonstrating exceptional performance with AUROCs of 0.9998 across multiple datasets. The method not only achieved high accuracy in identity retrieval but also effectively flagged identity inconsistencies, highlighting the robustness of retinal anatomy as a biometric signal across varying imaging conditions and extended follow-up periods.
Retinal biometrics can achieve near-perfect identity verification, with a Recall@1 of 99.7% across diverse imaging devices and patient demographics.
Patient identity errors can compromise longitudinal medical records, research databases, and downstream clinical decisions. We present a retinal biometric system for verifying claimed identities and retrieving the correct identity from color fundus images. We trained a 512-dimensional metric-learning encoder combining a ConvNeXtV2 backbone with ArcFace and triplet losses on 227,004 images from 21,851 patient-eye identities in the Rotterdam Study, spanning multiple imaging devices and up to 32.6 years of follow-up. The system was evaluated on held-out Rotterdam Study data and externally on the UK Biobank and Age-Related Eye Disease Study (AREDS). Before evaluation, we used the model to screen for identity inconsistencies and manually adjudicated flagged images, identifying incorrect assignments in 0.588% of Rotterdam Study images, 0.259% of UK Biobank images, and 0.164% of AREDS images. In retrospective-only verification after removing near-duplicate images, the system achieved AUROCs of 0.9998, 0.9997, and 0.9998 in the Rotterdam Study, UK Biobank, and AREDS, respectively. For identity retrieval using only previously acquired images, Recall@1 was 99.7%, 97.2%, and 97.6%, respectively, from galleries averaging 4436-8510 identities; the correct identity appeared among the top five results in at least 98.6% of cases. Performance remained robust across imaging devices and long follow-up intervals, while lower image quality and inconsistent retinal fields accounted for most failures. These findings establish retinal anatomy as a durable biometric signal, useful for safeguarding the integrity of longitudinal imaging records.