Retinal image analysis software for neurodegenerative disease research

我们的目标是开发和验证一个独立于设备的光学相干性分析软件应用程序 这是人类视网膜的断层扫描（OCT）图像。我们的系统将进行定量测量视网膜 黄斑处的层厚度，以支持用于测量发作的生物标志物的生成， 眼部和神经变性疾病的进展，包括年龄相关性黄斑变性（AMD）， 糖尿病性视网膜病变（DR）、青光眼、多发性硬化症（MS）、阿尔茨海默病、帕金森病和 肌萎缩侧索硬化症（ALS）。视网膜层厚度表明萎缩通过变薄和增加 液体或炎症通过增厚。精确、独立于设备的视网膜层分割 通过对层厚度的纵向分析可以返回许多定量生物标志物 与疾病发作和进展相关，并便于直接比较OCT器械与 正常，以估计异常的程度。具体而言，我们的目标是： 目标1 -改进我们对患病眼睛的分割 Orion（www.voxeleron.com/orion）是我们目前的研究平台，已经过验证，并在 正常的非病理性眼睛。然而，我们对视网膜层的结构分析可能会因视觉因素而变得复杂。 老年人群中普遍存在的疾病或混浊。目前的软件，包括我们的，在以下情况下性能很差： 疾病的情况下，我们的目标是改善我们目前的分割算法，在这些情况下， 病例，并在包括AMD、DR和青光眼的大型手动分割数据集上验证其性能 从至少4个不同器械制造商的OCT摄像机中获取的病例。 目标2 -将纵向分析添加到我们的细分软件中 在临床上，数据的静态分析具有有限的效用。我们将在现有的基础上增加纵向分析， 分割能力，以测量厚度随时间的变化。新的临床工作站将 通过确定与专家生成的地面实况的一致性来进行严格测试。

Our goal is to develop and validate a device-independent software application for analysis of optical coherence tomography (OCT) images of the human retina. Our system will make quantitative measurements of retinal layer thicknesses at the macula in support of the generation of biomarkers for measuring onset and progression of ocular and neurodegenerative diseases, including age-related macular degeneration (AMD), diabetic retinopathy (DR), glaucoma, multiple sclerosis (MS), Alzheimer’s disease, Parkinson’s disease, and amytrophic lateral sclerosis (ALS). Retinal layer thicknesses indicate atrophy through thinning and increased fluid or inflammation through thickening. Accurate, device-independent segmentation of retinal layers together with longitudinal analysis of the layer thicknesses can return a number of quantitative biomarkers to correlate with disease onset and progression, and facilitate direct comparison across OCT devices to results from normal to estimate the degree of abnormalities. Specifically, we are aiming to: Aim 1 – Improve our segmentation of diseased eyes Orion (www.voxeleron.com/orion), our current research platform has been validated and used extensively on normal, non-pathologic eyes. Our structural analyses of retinal layers may be complicated, however, by ocular diseases or opacities prevalent in an aging population. Current software, ours included, can perform poorly in the case of disease, a situation we aim to ameliorate by improving our current segmentation algorithm in these cases and validating its performance on a large, hand-segmented dataset including AMD, DR, and glaucoma cases taken from at least 4 different device manufacturers’ OCT cameras. Aim 2 – Add longitudinal analysis to our segmentation software Clinically, static analysis of data has limited utility. We will add longitudinal analysis to our existing segmentation capabilities to measure the change of thicknesses over time. This new clinical workstation will be rigorously tested by determining agreement with expert-generated ground-truth.