Dissemination of a Software Platform for Efficient CT Radiation Dose Optimization and Diagnostic Performance Assessment

项目概要/摘要 随着许多新技术的引入，以尽量减少 CT 中的辐射剂量，仍然存在很大的变化 CT 检查中规定的辐射剂量水平，因此诊断结果存在很大差异 性能。一些患者可能接受高于必要剂量的剂量。有些可能剂量不足或错误 诊断为图像质量不足。为了确定合适的辐射量 每次检查的剂量减少，需要准确量化诊断性能，以便剂量 可以在不牺牲重要诊断信息的情况下实现减少。然而，目前有一个 缺乏有效和定量的工具来客观评估诊断性能，特别是对于许多人来说 涉及数据非线性处理的新型剂量减少方法，例如迭代 基于重建和深度学习的降噪方法。 该应用程序的具体目标是传播高度自动化的解决方案，CT协议优化 (CTPro) 软件，面向广泛的 CT 社区。该定量工具提供了有效的实施 诊断性能评估和 CT 辐射剂量优化。该工具基于通道化 Hotelling观察者（CHO），它本身是几十年前开发的，用于模仿人类观察者的视觉反应 在信号检测任务中。然而，由于缺乏严格的标准，CHO在临床CT中的应用相当有限。 在实践中进行验证和高效、稳健的实施。我们是第一个证明其相关性的人 在临床 CT 的低对比度检测、分类和定位任务中具有人类观察者的性能。 当前提案的主要目标是优化该工具以实现简单性和稳健性，并且 将其传播给 CT 研究人员和临床用户，这将通过 3 个具体目标来实现： 目标 1：优化 CTPro 以实现简单性、稳健性和通用性。 目标 2：开发一个基于网络的开源软件传播平台。 目标 3：构建用例并传播 CTPro。 拟议的工作意义重大，因为该软件工具将允许任何 CT 用户和研究人员执行 CT 辐射剂量优化和诊断性能评估以高效、定量和准确的方式进行 客观的方式。这项工作的创新之处在于自动化工具将使用定量测量 诊断性能系统地指导 CT 剂量优化的复杂任务，超越 传统的指标不适用于许多新颖的剂量减少技术。软件工具，一次 广泛应用，将促进剂量优化和剂量减少评估的范式转变 技术已被执行，并将允许更快速和一致地采用剂量减少技术 进入临床，将造福数百万CT患者。

PROJECT SUMMARY/ABSTRACT With the introduction of many novel techniques to minimize radiation dose in CT, there is still a large variation in terms of radiation dose levels prescribed in CT exams and therefore a large variation of diagnostic performance. Some patients may receive higher dose than necessary. Some may be under-dosed and mis- diagnosed as a result of insufficient image quality. In order to determine the appropriate amount of radiation dose reduction in each exam, accurate quantification of diagnostic performance is needed so that the dose reduction can be achieved without sacrificing important diagnostic information. However, currently there is a lack of efficient and quantitative tools for objective assessment of diagnostic performance, particularly for many of the novel dose reduction methods involving non-linear processing of the data such as iterative reconstruction and deep-learning-based noise reduction methods. The specific goal of this application is to disseminate a highly automated solution, CT Protocol optimization (CTPro) software, to a wide CT community. This quantitative tool provides an efficient implementation of diagnostic performance assessment and CT radiation dose optimization. This tool is based on channelized Hotelling observer (CHO), which itself was developed decades ago to mimic human observer visual responses in signal detection tasks. However, the use of CHO in clinical CT is quite limited because of a lack of rigorous validation and efficient and robust implementation in practice. We were the first to demonstrate its correlation with human observer performance in low-contrast detection, classification and localization tasks in clinical CT. The main objective of the current proposal is to optimize this tool for simplicity and robustness, and disseminate it to CT researchers and clinical users, which will be accomplished through 3 specific aims: Aim 1: Optimize CTPro for simplicity, robustness, and generalizability. Aim 2: Develop an open-source web-based platform for software dissemination. Aim 3: Build use cases and disseminate CTPro. The proposed work is significant because the software tool will allow any CT users and researchers to perform CT radiation dose optimization and diagnostic performance evaluation in an efficient, quantitative, and objective manner. This work is innovative in that the automated tool will use quantitative measures of diagnostic performance to systematically guide the complex task of CT dose optimization, moving beyond traditional metrics that are inappropriate for many novel dose reduction techniques. The software tool, once widely employed, will facilitate a paradigm shift in how dose optimization and the evaluation of dose reduction techniques are performed, and will allow a more rapid and consistent adoption of dose reduction technology into clinical practice, which will benefit millions of CT patients.