Woof William A, de Guimarães Thales A C, Al-Khuzaei Saoud, Daich Varela Malena, Sen Sagnik, Bagga Pallavi, Mendes Bernardo, Shah Mital, Burke Paula, Parry David, Lin Siying, Naik Gunjan, Ghoshal Biraja, Liefers Bart J, Fu Dun Jack, Georgiou Michalis, Nguyen Quang, Sousa da Silva Alan, Liu Yichen, Fujinami-Yokokawa Yu, Sumodhee Dayyanah, Patel Praveen, Furman Jennifer, Moghul Ismail, Moosajee Mariya, Sallum Juliana, De Silva Samantha R, Lorenz Birgit, Holz Frank G, Fujinami Kaoru, Webster Andrew R, Mahroo Omar A, Downes Susan M, Madhusudhan Savita, Balaskas Konstantinos, Michaelides Michel, Pontikos Nikolas
University College London Institute of Ophthalmology, London, United Kingdom.
Moorfields Eye Hospital NHS Foundation Trust, London, United Kingdom.
Ophthalmol Sci. 2024 Nov 12;5(2):100652. doi: 10.1016/j.xops.2024.100652. eCollection 2025 Mar-Apr.
To quantify relevant fundus autofluorescence (FAF) features cross-sectionally and longitudinally in a large cohort of patients with inherited retinal diseases (IRDs).
Retrospective study of imaging data.
Patients with a clinical and molecularly confirmed diagnosis of IRD who have undergone 55° FAF imaging at Moorfields Eye Hospital (MEH) and the Royal Liverpool Hospital between 2004 and 2019.
Five FAF features of interest were defined: vessels, optic disc, perimacular ring of increased signal (ring), relative hypo-autofluorescence (hypo-AF), and hyper-autofluorescence (hyper-AF). Features were manually annotated by 6 graders in a subset of patients based on a defined grading protocol to produce segmentation masks to train an artificial intelligence model, AIRDetect, which was then applied to the entire imaging data set.
Quantitative FAF features, including area and vessel metrics, were analyzed cross-sectionally by gene and age, and longitudinally. AIRDetect feature segmentation and detection were validated with Dice score and precision/recall, respectively.
A total of 45 749 FAF images from 3606 patients with IRD from MEH covering 170 genes were automatically segmented using AIRDetect. Model-grader Dice scores for the disc, hypo-AF, hyper-AF, ring, and vessels were, respectively, 0.86, 0.72, 0.69, 0.68, and 0.65. Across patients at presentation, the 5 genes with the largest hypo-AF areas were , , , , and , with mean per-patient areas of 43.72, 29.57, 20.07, 19.65, and 16.92 mm, respectively. The 5 genes with the largest hyper-AF areas were , , , , and , with mean areas of 0.50, 047, 0.44, 0.38, and 0.33 mm, respectively. The 5 genes with the largest ring areas were , , , and , with mean areas of 3.60, 2.90, 2.89, 2.56, and 2.20 mm, respectively. Vessel density was found to be highest in , , , , and (11.0%, 10.4%, 10.1%, 10.1%, 9.2%) and was lower in retinitis pigmentosa (RP) and Leber congenital amaurosis genes. Longitudinal analysis of decreasing ring area in 4 RP genes (, , , and ) found to be the fastest progressor at -0.178 mm/year.
We have conducted the first large-scale cross-sectional and longitudinal quantitative analysis of FAF features across a diverse range of IRDs using a novel AI approach.
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
对大量遗传性视网膜疾病（IRD）患者进行眼底自发荧光（FAF）相关特征的横断面和纵向定量分析。
成像数据的回顾性研究。
2004年至2019年间在摩尔菲尔德眼科医院（MEH）和皇家利物浦医院接受55°FAF成像、临床和分子诊断确诊为IRD的患者。
定义了五个感兴趣的FAF特征：血管、视盘、信号增强的黄斑周围环（环）、相对低自发荧光（低AF）和高自发荧光（高AF）。6名分级人员根据定义的分级方案在一部分患者中手动标注特征，以生成分割掩码，用于训练人工智能模型AIRDetect，然后将其应用于整个成像数据集。
通过基因和年龄对定量FAF特征（包括面积和血管指标）进行横断面分析和纵向分析。分别用Dice分数和精确率/召回率对AIRDetect特征分割和检测进行验证。
使用AIRDetect自动分割了来自MEH的3606例IRD患者的45749张FAF图像，涵盖170个基因。视盘、低AF、高AF、环和血管的模型-分级人员Dice分数分别为0.86、0.72、0.69、0.68和0.65。在初诊患者中，低AF面积最大的5个基因是 、 、 、 和 ，每位患者的平均面积分别为43.72、29.57、20.07、19.65和16.92平方毫米。高AF面积最大的5个基因是 、 、 、 和 ，平均面积分别为0.50、0.47、0.44、0.38和0.33平方毫米。环面积最大的5个基因是 、 、 、 和 ，平均面积分别为3.60, 2.90, 2.89, 2.56和2.20平方毫米。发现血管密度在 、 、 、 和 中最高（11.0%、10.4%、10.1%、10.1%、9.2%），在视网膜色素变性（RP）和莱伯先天性黑蒙基因中较低。对4个RP基因（ 、 、 、 和 ）中环面积减小的纵向分析发现， 是进展最快的，每年减小0.178平方毫米。
我们使用一种新颖的人工智能方法对多种IRD的FAF特征进行了首次大规模横断面和纵向定量分析。
在本文末尾的脚注和披露中可能会发现专有或商业披露信息。
