Validating Automated Photoreceptor Analytics Software For Degenerative Eye Disease Research and Biopharma Clinical Trials.

摘要 神经变性涉及几乎所有视网膜视力丧失的主要原因，特别是遗传性视网膜病变。 退化性眼病虽然视觉功能测试和主要的成像方式，如光学相干断层扫描， 对于有症状的患者来说，是受欢迎的诊断和治疗管理工具，但这些临床工具并不令人满意 作为判断新的预防性和恢复性治疗的临床疗效的预后指标或终点（例如， 神经保护剂、基因疗法或干细胞疗法），其在视网膜中的细胞水平上起作用。自适应光学 (AO)检眼镜检查已经成为光感受器的存在和活力的灵敏标记;然而， 没有基于AO眼底图像的经验证的算法或客观定量测量， 代表性的图像数据库，基于该图像数据库进行筛选判断。在本直接至阶段II SBIR中， 成像创新（TII）和约瑟夫卡罗尔教授，高级眼科成像计划（AOIP）主任，医学 威斯康星州学院（MCW）将验证和商业化自动感光器分析软件， 退行性眼病。 该平台将利用Mosaic Analytics（MOSAIC），这是一种自动感光器分析软件包， MCW和AOIP图像库，其中包含1578个主题的图像和数据-336个正常人和其余的 患有一百种视网膜疾病中的一种或多种MOSAIC将释放AO增强检眼镜的潜在价值 提供可靠、客观、直接的光感受器健康测量，并提供 评估细胞疗法的临床疗效。为了实现这一目标，我们将提出四个目标： 通过先验图像质量评估和后验置信度评估的感光器处理算法， 采用感兴趣的减法区域，并且算法调谐到感兴趣的视网膜域;（B）将感兴趣的减法区域与感兴趣的视网膜域进行比较; 用于计算客观定量生物标志物的阵列的算法;（c）建立适当的使用背景， 在FDA MDDT计划中对客观临床试验终点进行资格鉴定;以及（d）发布第一个规范性文件 健康/患病眼睛中定量光感受器生物标志物的参考数据库。 我们的提案填补了视网膜成像领域的一个重要技术空白。虽然成像的数量和类型 设备持续增长，但用于评估和管理这些设备中的映像的分析工具尚未开发出来， 并联因此，这些精致的成像设备的诊断潜力仍未实现。的验证 定量自适应光学生物标志物将增加临床试验结果的信心，并减少 新的细胞疗法的市场。MOSAIC的可靠性、可重复性和易用性将促进 自适应光学眼底成像和加速致盲性退行性疾病治疗的发展。

ABSTRACT Neurodegeneration is implicated in almost all major causes of retinal vision loss and specifically in inherited degenerative eye disease. While visual function tests and major imaging modalities, such optical coherence tomography, are favored diagnostic and treatment management tools for symptomatic patients, these clinical tools are unsatisfactory as prognostic indicators or as endpoints for judging clinical efficacy of new preventative and restorative therapies (e.g., neuro-protectives, gene therapies, or stem cell therapies) that operate at the cellular level in the retina. Adaptive optics (AO) ophthalmoscopy has emerged as a sensitive marker of the presence and viability of photoreceptors; however, there are no validated algorithms or objective quantitative measures based on AO fundus images and there are no representative image databases upon which to base screening judgements. In this Direct-to-Phase II SBIR, Translational Imaging Innovations (TII) and Prof. Joseph Carroll, Director of the Advanced Ocular Imaging Program (AOIP), Medical College of Wisconsin (MCW), will validate and commercialize an Automated Photoreceptor Analytics Software for Degenerative Eye Disease. This platform will leverage Mosaic Analytics (MOSAIC), an automated photoreceptor analysis package developed at MCW, and the AOIP Image Bank, which houses images and data on 1578 subjects - 336 normals and the remainder afflicted with one or more of 100 retinal diseases. MOSAIC will unlock the latent value of AO-enhanced ophthalmoscopy to provide a reliable, objective, direct measure of photoreceptor health, and provide quantitative endpoints for assessing the clinical efficacy of cellular therapies. To achieve this goal, we will propose four aims: (a) Strengthen the Photoreceptor Processing Algorithm(s) through a priori image Quality Metrics and a posteriori Confidence Metrics, adoption of Subtractive Regions of Interest, and algorithm tuning to retinal Domains of Interest; (b) Validate the Algorithm(s) for Computing an array of Objective Quantitative Biomarkers; (c) Establish a Proper Context of Use for Qualifying an Objective Clinical Trial Endpoint within the FDA MDDT program; and (d) Publish the first Normative Reference Database for Quantitative Photoreceptor Biomarkers in healthy/diseased eyes. Our proposal fills an important technology gap in the field of retinal imaging. While the number and type of imaging devices continues to grow, the analytical tools to assess and manage images from these devices have not developed in parallel. As such, the diagnostic potential of these exquisite imaging devices remains unrealized. The validation of quantitative adaptive optics biomarkers will increase confidence in the outcome of clinical trials and reduce time to market for new cellular therapies. The reliability, reproducibility, and ease of use of MOSAIC will catalyze the adoption of adaptive optics fundus imaging and accelerated the development of therapies for blinding degenerative diseases.