Monitoring of Glaucoma Patients in Advanced Disease

晚期青光眼患者的监测

• 批准号: 10503781
• 负责人: Sasan Moghimi Araghi
• 金额: $ 51.56万
• 依托单位: UNIVERSITY OF CALIFORNIA, SAN DIEGO
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-09-01 至 2027-08-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10503781
• 关键词: Address; Affect; Algorithms; Angiography; Area; Artificial Intelligence; Blindness; Blood Vessels; Caring; Central Scotomas; Classification; Clinical; Cluster Analysis; Complex; Data; Databases; Defect; Detection; Deterioration; Development; Diagnostic; Disease; Dropout; Early Diagnosis; Economic Burden; Evaluation; Eye; Floor; Future; Ganglion Cell Layer; Glaucoma; Goals; Image; Imaging technology; Individual; Island; Knowledge; Machine Learning; Mathematics; Measurement; Measures; Methods; Modality; Modeling; Monitor; Nerve Degeneration; Onset of illness; Open-Angle Glaucoma; Optic Disk; Optical Coherence Tomography; Patients; Pattern; Pattern Recognition; Public Health; Quality of life; Residual state; Retina; Risk; Risk Assessment; Series; Statistical Methods; Structure; Structure-Activity Relationship; Supervision; System; Techniques; Testing; Texture; Thick; Thinness; Time; United States National Institutes of Health; Vision; Visual Fields; Visual Pathways; Visual impairment; Work; advanced disease; aggressive therapy; archetypal analysis; artificial intelligence algorithm; base; burden of illness; care providers; central visual field; clinical care; clinical decision-making; cost; deep learning; deep learning algorithm; density; disability; field study; follow-up; functional status; ganglion cell; high risk; improved; machine learning model; macula; multimodality; new technology; novel; optic nerve disorder; optical imaging; patient oriented; prediction algorithm; programs; retinal nerve fiber layer

Project Summary Individuals with advanced glaucomatous damage have markedly impaired visual function resulting in a decreased quality of life. This proposal will provide the longitudinal follow-up to fill in important gaps in our knowledge about monitoring eyes with advanced open angle glaucoma (OAG). Monitoring of the disease in its advanced stages is challenging because the visual field island shrinks to such an extent that only the central visual field survives and measurement of the retinal nerve fiber layer thickness reaches a floor, after which more thinning is not detectable. The overall objectives of this application are (i) to characterize the macular structural (microvasculature and thickness) and functional changes in eyes with advanced OAG and (ii) to develop novel models that can detect and predict progression in these eyes. The central hypothesis is that novel statistical and artificial intelligence-based analyses of central visual field functional status and recently developed macular optical imaging measurements will improve monitoring of disease in advanced OAG eyes. There is a critical need for models that can predict glaucomatous progression in advanced OAG eyes and to characterize longitudinal loss of macular structure and function in order to advise clinical decision making. The central hypothesis will be tested by 3 Specific Aims. Aims 1 and 2 will develop and validate models for detection of OAG progression using the central 10 degree visual field and characterize patterns of the longitudinal changes in the central visual field and retina. Cluster-based progression methods will be applied in vulnerable and less vulnerable zones of the 10-2 visual field. Nested multivariable linear mixed effects models will be used to compare rates of macula structure (ganglion cell layer and vessel density) and functional change (in eyes with Mean Deviation <-8 dB) and to characterize the relationships between baseline patterns of visual field and structural loss and glaucoma progression while adjusting for inter-eye correlation. In Aim 3, we will apply novel deep learning techniques to macular function and recently developed optical imaging measurements to improve the prediction accuracy of glaucomatous progression in advanced disease. Complex functional and structural tests in daily use by eye care providers contain hidden information that is not fully used in the current analyses and advanced pattern recognition/machine learning-based analysis techniques can find and use that hidden information. We will use mathematically rigorous unsupervised techniques such as archetypal analysis and multimodal deep learning to discover patterns of defects and assess the risk of changes in longitudinal series of perimetric and optical imaging data from >500 patients, available in our NIH-supported glaucoma database. The proposed work is significant because it will lead to development of more effective mathematically-based, validated methods of detecting OAG progression in eyes with advanced disease. Moreover, it will reduce the cost of glaucoma care by identifying high-risk patients that require more aggressive treatment, thus decreasing disability and reducing the burden of glaucoma blindness.

项目概要 患有晚期青光眼损伤的个体视觉功能明显受损，导致 生活质量下降。该提案将提供纵向后续行动，以填补我们的重要空白 有关监测晚期开角型青光眼 (OAG) 眼睛的知识。监测其疾病状况 高级阶段具有挑战性，因为视野岛缩小到只有中央区域 视野得以保留，视网膜神经纤维层厚度的测量达到底部，之后 更多的稀疏是无法检测到的。该应用的总体目标是 (i) 表征黄斑 患有晚期 OAG 的眼睛的结构（微血管和厚度）和功能变化以及 (ii) 开发可以检测和预测这些眼睛的进展的新模型。中心假设是 基于新颖的统计和人工智能的中央视野功能状态分析以及最近 开发的黄斑光学成像测量将改善对晚期 OAG 眼疾病的监测。 迫切需要能够预测晚期 OAG 眼青光眼进展的模型 表征黄斑结构和功能的纵向损失，以便为临床决策提供建议。这 中心假设将通过 3 个具体目标进行检验。目标 1 和 2 将开发和验证模型 使用中央 10 度视野检测 OAG 进展并表征 OAG 的模式 中央视野和视网膜的纵向变化。基于聚类的进展方法将应用于 10-2 视野的脆弱区域和较不脆弱区域。嵌套多变量线性混合效应模型 将用于比较黄斑结构（神经节细胞层和血管密度）和功能的比率 变化（在平均偏差 <-8 dB 的情况下）并表征基线模式之间的关系 视野和结构损失以及青光眼进展，同时调整眼间相关性。在目标 3 中， 我们将把新颖的深度学习技术应用于黄斑功能和最近开发的光学成像 测量以提高晚期疾病青光眼进展的预测准确性。 眼保健提供者日常使用的复杂功能和结构测试包含隐藏的信息，这些信息不是 完全用于当前的分析和基于高级模式识别/机器学习的分析 技术可以找到并使用隐藏的信息。我们将使用数学上严格的无监督 原型分析和多模态深度学习等技术来发现缺陷模式和 评估超过 500 名患者的纵向系列视野和光学成像数据变化的风险， 可在我们 NIH 支持的青光眼数据库中找到。拟议的工作意义重大，因为它将导致 开发更有效的基于数学的、经过验证的方法来检测眼部 OAG 进展 患有晚期疾病。此外，它将通过识别高危患者来降低青光眼护理成本 需要更积极的治疗，从而减少残疾并减轻青光眼失明的负担。