STEREOSCOPIC GRADING SYSTEM--AGE RELATED MACULOPATHY

Kestrel提出了一种使用可负担得起的立体可视化设备和现代计算机接口的年龄相关性黄斑病变（ARMD）立体分级系统。我们的目标是开发一种基于计算机的系统，用于对显示体积病变的视网膜图像进行分级，并在减少阅读器可变性的同时提高效率和定量。技术的革命性进步使人们有可能将现有的“光桌和手持立体观众”移到电脑屏幕上。该系统将通过降低对大量患者数据进行分级的成本而产生立竿见影的影响。最终，拟议的系统将成为一个重要因素，通过它在试图确定ARMD因果影响的流行病学研究中的应用。Kestrel的研究方法包括：(1)使用数字化立体图像中的颜色数据自动进行分级活动，同时将计数和测量病变的负担从分级员转移到计算机上。(2)通过生成地形图来分析立体照片，并将地形图输入神经网络，该神经网络将检测和分类armd相关病理。创新包括：(1)显式三维视差图生成。(2)目前的3d技术作为人工分级的替代方案。(3)指令和数据录入语音识别接口。(4)神经网络算法增强了特征分割和分类。开发用于视网膜图像分级的商业立体系统，可以对病理进行精确测量和量化，以进行统计分析。该系统将通过降低对大量患者图像进行分级的成本，对流行病学研究产生直接影响。作为确定ARMD治疗新方法疗效的定量工具，ARMD立体分级系统将提供给药品和研究人员。

Kestrel proposes a Stereoscopic Grading System for Age-related Maculopathy (ARMD) using affordable stereoscopic visualization devices and modern computer interfaces. Our objective is to develop a computer-based system for grading retinal images which display volumetric lesions, and to increase efficiency and quantitativeness while reducing reader variability. Revolutionary advances in technology make it possible to move exiting "light-table and hand-held stereoscopic viewers" to the computer screen. This system will have immediate impact by reducing the cost of grading large volumes of patient data. Ultimately, the proposed system will be a significant factor, through its application in epidemiologic studies which attempt to identify causal effects of ARMD. Kestrel's research approach involves (1) Automating grading activity using color data from digitized stereoscopic images, while moving the burden of counting and measuring lesions from the grader to the computer. (2) Analyzing stereoscopic photographs by generating topographical maps for input into a neural network which will detect and classify ARMD-related pathologies. The innovations include: (1) Explicit 3-D disparity maps generation. (2) Current 3-D technology as an alternative to manual grading. (3) Speech recognition interface for instruction sand data entry. (4) Neural network algorithms enhance feature segmentation and classification. PROPOSED COMMERCIAL APPLICATION Development of a commercial stereoscopic system for grading retinal images allows the precise measurement and quantification of pathologies for statistical analyses. The system will have an immediate impact on epidemiological studies by reducing the cost of grading large populations of patient images. As a quantitative tool for determining the efficacy of new procedures for treating ARMD, the Stereoscopic Grading System for ARMD will be offered to pharmaceuticals and researchers.