德鲁森
视盘
验光服务
眼科
计算机科学
人工智能
医学
黄斑变性
视网膜
作者
Yan Yan,Cassie A. Ludwig,Yaping Joyce Liao
标识
DOI:10.1016/j.ajo.2020.12.023
摘要
To identify key en face multimodal imaging features of optic disc drusen (ODD).Retrospective cross-sectional study.Setting: a single academic center. Patient orStudyPopulation: 786 patients (10-82 years of age) with diagnostic codes for optic disc drusen (ODD) in clinical notes extracted using natural language processing. Intervention orObservationProcedures: color fundus imaging, green-light and blue-light fundus autofluorescence (FAF), near-infrared reflectance (NIR), and enhanced-depth imaging optical coherence tomography (EDI-OCT). MainOutcomeMeasurements: Ophthalmic imaging characteristics and sensitivity of en face imaging compared with EDI-OCT.A total of 38 patients (61 eyes) had high-quality EDI-OCT scans and en face multimodal imaging. Green-light FAF imaging had the highest diagnostic sensitivity (96.8%) for ODD and showed homogeneous hyperautofluorescence, whereas blue-light FAF imaging had heterogeneous brightness, which helped differentiate superficial from deep ODD. Blue-light FAF (93.5%) and NIR (91.8%) imaging were also sensitive tests and revealed papillary and peripapillary features that were not well seen on green-light FAF, including the size and depth of ODD, morphology of the optic disc, and common ODD-associated structures such as horizontal hyper-reflective lines and peripapillary hyper-reflective ovoid mass-like structures (PHOMS). Color fundus imaging had the lowest sensitivity (82%). There was good inter-rater reliability for all en face imaging modalities (P < .0001 for all).In en face imaging, green-light FAF had the highest sensitivity for the diagnosis of ODD, whereas blue-light FAF and NIR images provided more information regarding the severity, location, depth, and size of ODD. In eyes that are negative on green-light FAF, EDI-OCT should be performed and provides the highest-resolution characterization of the entire optic disc to assess or rule out ODD.
科研通智能强力驱动
Strongly Powered by AbleSci AI