Real-time, Automatic Image Quality Assessment for Digital Fundus Camera

数码眼底相机的实时、自动图像质量评估

• 批准号: 8323412
• 负责人: MARIOS S PATTICHIS
• 金额: $ 34.79万
• 依托单位: VISIONQUEST BIOMEDICAL, LLC
• 依托单位国家: 美国
• 财政年份: 2010
• 资助国家: 美国
• 起止时间: 2010-12-01 至 2013-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/8323412
• 关键词: Address; Algorithms; Applications Grants; Architecture; Award; Back; Characteristics; Clinic; Collaborations; Color; Computer software; Data; Databases; Detection; Development; Diabetes Mellitus; Diagnosis; Diagnostic; Elements; Ensure; Environment; Feedback; Funding; Fundus; Healthcare; Human; Image; Imagery; Individual; Institutes; Iowa; Journals; Legal patent; Location; Manufacturer Name; Methodology; Methods; Metric; Modeling; Morphologic artifacts; Mydriatics; New Mexico; Ophthalmologist; Ophthalmology; Paper; Patients; Perception; Pharmaceutical Preparations; Phase; Photography; Positioning Attribute; Prospective Studies; Protocols documentation; Publishing; Reading; Recommendation; Reporting; Research; Retina; Retinal; Retinal Diseases; Screening procedure; Societies; Source; South Texas; Specific qualifier value; Specificity; Spottings; Study Subject; System; Techniques; Testing; Texas; Time; Training; Universities; Validation; Visual Pathways; Wisconsin; base; clinical research site; computerized data processing; digital; digital imaging; health care quality; interest; meetings; millisecond; operation; professor; prospective

DESCRIPTION (provided by applicant): Real-time image quality is a critical requirement in a number of healthcare environments. Additionally, non-real-time applications, such as research and drug studies suffer loss of data due to unusable (untradeable) retinal images. Several published reports indicate that from 10% to 15% of images are rejected from studies due to image quality. With the transition of retinal photography to lesser trained individuals in clinics, image quality may suffer unless there is a means to assess the quality of an image in real-time and give the photographer recommendations for correcting technical errors in the acquisition of the photograph. In Phase I, this project demonstrated a methodology for evaluating a digital image from a funds camera in real-time and giving the operator feedback as to the quality of the image. We showed that it is possible to identify the source of the problem in poor quality images and give the photographer corrective actions. By providing real-time feedback to the photographer, corrective actions can be taken and loss of data or inconvenience to the patient eliminated. We successfully applied our methodology to over 2,000 images from four different cameras under mydriatic and non-mydriatic imaging conditions. We showed that the technique was equally effective on uncompressed and compressed (JPEG) images. In Phase II, we will validate the methodology further on additional data with different characteristics to demonstrate its broad applicability. Because our methodology uses parameters that are suggested by human perception qualities, we have shown that the algorithm can adapt to a variety of image quality protocols. The real- time retinal image quality methodology is based on image quality scores assigned by graders or ophthalmologists. Commercially, a real-time image quality assessment system is of interest to many manufacturers of fundus cameras. Our methodology will be demonstrated to be scalable to any digital imagery. We will integrate the algorithm into the image acquisition software of two commercial cameras (Topcon and Canon). Our methodology will also be of great value to screening centers where poor quality images can be reported immediately to the local or remote photographer. Commercially there will be three products: One, we will integrate the software directly into fundus cameras' image acquisition software. Two, we will produce a stand-alone image quality software package for use by individuals in clinics or research. Three, we will integrate our software and adapt it to specific protocols, such as the Wisconsin Fundus Photo Reading Center. 2

描述(由申请人提供)：实时图像质量是许多医疗保健环境的关键要求。此外，非实时应用，如研究和药物研究，由于不可用(不可交易)的视网膜图像而导致数据丢失。一些已发表的报告表明，有10%到15%的图像由于图像质量的原因而被研究拒绝。随着视网膜摄影向临床上训练有素的人过渡，图像质量可能会受到影响，除非有一种方法可以实时评估图像的质量，并向摄影师提供纠正照片获取中的技术错误的建议。在第一阶段，该项目演示了一种方法，用于实时评估Funds相机的数字图像，并就图像质量向操作员提供反馈。我们展示了在劣质图像中找出问题的根源并给摄影师纠正措施是可能的。通过向摄影师提供实时反馈，可以采取纠正措施，消除数据丢失或给患者带来的不便。我们成功地将我们的方法应用于散瞳和非散瞳成像条件下来自四个不同相机的2000多张图像。我们展示了该技术在未压缩和压缩(JPEG)图像上同样有效。在第二阶段，我们将在更多具有不同特征的数据上进一步验证该方法，以证明其广泛的适用性。由于我们的方法使用了人类感知质量所建议的参数，因此我们已经证明了该算法可以适应各种图像质量协议。实时视网膜图像质量方法是基于评分员或眼科医生分配的图像质量分数。在商业上，许多眼底相机制造商都对实时图像质量评估系统感兴趣。我们的方法将被证明可以扩展到任何数字图像。我们将该算法集成到两款商用相机(Topcon和佳能)的图像采集软件中。我们的方法对筛选中心也很有价值，在那里，质量不佳的图像可以立即报告给本地或远程摄影师。商业上将有三个产品：第一，我们将把软件直接集成到眼底相机的图像采集软件中。第二，我们将制作一个独立的图像质量软件包，供个人在临床或研究中使用。第三，我们将整合我们的软件，并使其适应特定的协议，如威斯康星州眼底照片阅读中心。2.