TOPIC #411 - PHASE I SBIR CONTRACT - IMAGEDEPHI: IMAGE DE-IDENTIFICATION OF OR THE ACCELERATION OF CANCER RESEARCH

该项目将产生一种统一的方法，用于检测、匿名化和审查成像数据文件中受保护的健康信息（PHI），特别是在整个幻灯片图像中发现的PHI。为了共享癌症研究的成像数据，数据文件需要通过删除所有PHI而不删除或降级数据来匿名化。将使用一组可扩展的文件格式处理程序来读取各种各样的图像数据以及相关的文本信息。将开发一套算法来检测文本和图像中的PHI。图像和文本将被去识别，在跟踪来源的同时修改原始文件。将开发一个用户界面，方便审查去识别结果，使研究人员能够确认匿名化，并进一步处理未完全去识别的数据文件。在开发检测、修改和评审过程中，将测量结果的质量，允许每个步骤随着时间的推移得到改进。这将减轻研究人员之间共享成像数据的负担，允许更广泛的信息用于研究和临床工作。

This project will produce a unified method for detection, anonymization, and review of Protected Health Information (PHI) in imaging data files, especially as found in whole slide images. In order to share imaging data for cancer research, data files need to be anonymized by removing all PHI without removing or degrading the data more than necessary. An expandable set of file format handlers will be used to read a wide variety of image data along with associated textual information. A set of algorithms will be developed that detect PHI in text and images. Images and text will be de-identified, modifying the original files while tracking provenance. A user interface will be developed to easily facilitate review of the de-identified results, enabling a researcher to confirm anonymization and to further process data files that were not fully deidentified. In developing the detection, modification, and review process, the quality of the results will be measured, allowing each step to be improved over time. This will reduce the burden of sharing imaging data between researchers, allowing a broader range of information to be used in research and clinical work.