Generation and Dissemination of Enhanced AI/ML-ready Prostate Cancer Imaging Datasets for Public Use

项目摘要 父母补助金汇总表 先进技术-国家图像引导治疗中心（AT-NCIGT）是一个研究和 技术中心，其使命是通过开发新颖的创新工具， 指导治疗（IGT）。这三项技术包括成像癌症异质结，深度学习和 将在单独和交叉TR&D中研究用于图像引导治疗的术中器械 组合。母公司AT-NCIGT项目包括开发开源平台，以支持 管理和分析在图像引导程序期间收集的成像数据。这个平台正在 在TRD 2下的目标之一中开发。本补编下拟议的工作属于 具体来说，AT-NCIGT深度学习TRD正在进行的工作，因为它提出开发技术， 将支持为AT-NCIGT -前列腺癌的优先级应用程序收集的数据集， 元数据，使其AI/ML准备就绪。为协调和丰富前列腺癌MRI而开发的工具 数据集将支持平台与最新的AI/ML分析工具及其评估的集成， 项目调查人员和外部合作者，而公开的策划数据集发布 作为补充的可交付成果，将支持前列腺癌成像中AI/ML工具的外部开发。 补充项目的目标 我们建议通过执行以下操作，使前列腺癌成像数据更适合AI/ML，1） 自动生成关于MRI扫描类型的关键缺失元数据，包括 模态、造影剂的使用和直肠内线圈的存在，2）增强公共数据集和BWH数据集 使用此元数据并共享结果，以及3）演示如何在AI/ML中使用增强的数据 申请和分发教材。通过这个项目，我们将能够提供公开的 以标准化格式丰富的数据集，适用于AI和ML开发方法， 前列腺癌的检测和风险评估。

PROJECT SUMMARY Summary of parent grant The Advanced Technologies-National Center for Image guided Therapy (AT-NCIGT) is a research and technology center with the mission of advancing patient care, by developing novel innovative tools for image guided Therapy (IGT). The three technologies encompass Imaging Cancer Heterogeneity, Deep learning and Intraoperative devices for image guided therapy which will be investigated both individually and in cross TR&D combinations. The parent AT-NCIGT project includes development of the open source platform to support management and analysis of imaging data collected during image-guided procedures. This platform is being developed in one of the aims under the TRD2. The work proposed under this supplement is within the scope of work being done in AT-NCIGT Deep Learning TRD specifically, since it is proposing to develop technology that will support enriching the datasets collected for an application of priority for AT-NCIGT - prostate cancer - with the metadata to make it AI/ML ready. The tools developed for harmonizing and enriching prostate cancer MRI datasets will support integration of the platform with the latest AI/ML analysis tools and their evaluation, both by the project investigators and by the external collaborators, while the publicly available curated dataset released as a deliverable of the supplement will support external development of AI/ML tools in prostate cancer imaging. Goals of the supplement project We propose to make prostate cancer imaging data more AI/ML ready by performing the following, 1) Automatically generate critical missing metadata regarding the type of MRI scan perform including the modality, use of contrast, and presence of an endorectal coil, 2) Enhance public datasets and a BWH dataset with this metadata and share the results, and 3) Demonstrate how to use the enhanced data in an AI/ML application and distribute teaching materials. Through this project, we will be able to provide publicly available enriched datasets in standardized formats that are suitable for AI and ML development of methods for the detection and risk assessment of prostate cancer.