INTERACTIVE COMPUTER-AIDED DIAGNOSIS TOOLS FOR GROUND-GLASS OPACITY LUNG TUMORS

这个子项目是许多研究子项目中利用 资源由NIH/NCRR资助的中心拨款提供。子项目和 调查员(PI)可能从NIH的另一个来源获得了主要资金， 并因此可以在其他清晰的条目中表示。列出的机构是 该中心不一定是调查人员的机构。 高分辨率计算机断层扫描(HRCT)经常被用来检测患者的肿瘤，并在治疗期间的不同时间间隔监测肿瘤的生长或缩小。准确地将肿瘤分为良性或恶性类别对于确定适当的治疗至关重要，而CT通常用于评估所选治疗的有效性。CT成像技术的进步有助于获得分辨率越来越高的图像；然而，目前的算法仅限于测量肿瘤的体积变化，而不能提供三维精确的肿瘤生长测量。这项研究特别感兴趣的是磨玻璃不透明(GGO)肿瘤，它对传统的图像分析算法构成了特殊的挑战，传统的图像分析算法倾向于检测高梯度变化，因此经常会错过GGO肿瘤。磨玻璃混浊是指在高分辨率计算机断层扫描(HRCT)过程中出现的模糊混浊，它不会遮盖 相关的肺血管。这种表现是由于脑实质的异常 均低于HRCT的空间分辨率。 在这项研究中，我们开发了一种新的三维(3D)方法，用于交互式、自动化和准确的GGO肿瘤分割和评估。我们方法的创新之处在于开发了新的交互式3D图像分析工具来提取GGO肺结节，并基于生成的不透明图进行分析。 到目前为止，现有的软件算法能够帮助检测和测量实体肺结节 基于可用的CT图像信息；但是，他们不能在GGO上工作 并估计检测到的肺结节的总体GGO覆盖率。目前的方法利用人工专家分析来完成这一重要任务。我们建议从CT图像中定量测量磨玻璃状不透明肿瘤中每个像素的不透明特性。我们的方法产生一个不透明贴图，其中每个像素的不透明度值都在0-1之间。对于给定的CT图像，我们建议通过构造图拉普拉斯矩阵和解线性方程组来完成估计，并辅以一些手工绘制的涂鸦，这些涂鸦的不透明度值很容易手工确定。 GGO肺肿瘤自动检测系统的开发将极大地提高常规放射和肿瘤学分析的效率。我们对GGO肿瘤进行客观评估的创新方法将使放射科医生或胸科医生能够评估肿瘤的三维演变以及在不同时间跨度进行的CT扫描检测到的维度变化，包括生长模式、最大生长面积/生长方向和不透明变化。这项拟议的研究是开发GGO肿瘤计算机评估的第一步，如果成功，将导致进一步的翻译努力，将这些技术整合到临床实践中。为成功开展这项工作而组建的团队包括一名胸外科医生，他担任临床主题专家，以及在图像增强、自动视觉和生物医学成像方面经验丰富的研究人员。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. High-resolution Computed Tomography (HRCT) is frequently used to detect tumors in patients, and to monitor tumor growth or shrinkage at different time intervals during treatment. The accurate classification of a tumor into benign or malignant categories is critical to determine the appropriate treatment and CTs are often used to assess the effectiveness of a selected treatment. Advances in CT imaging technology have assisted in acquiring the images at increasingly high resolution; however, current algorithms are limited to measuring volume changes of the tumor rather than providing an accurate measurement of tumor growth in three dimensions. Of particular interest for this study are Ground-Glass Opacity (GGO) tumors that pose a special challenge to conventional image analysis algorithms, which are traditionally tuned toward detection of high gradient changes and thus would frequently miss GGO tumors. Ground-glass opacity refers to the appearance of a hazy opacity during high-resolution computed tomography (HRCT) that does not obscure the associated pulmonary vessels. This appearance results from parenchymal abnormalities that are below the spatial resolution of HRCT. In this study, we develop a novel three-dimensional (3D) method for interactive, automated and accurate segmentation and assessment of GGO tumors. The innovation of our method is the development of novel interactive 3D image analysis tool to extract GGO lung nodules, and perform analysis based on the resulting opacity map. To date, existing software algorithms are able to help detect and measure solid lung nodules based on available CT-image information; however, they are not capable of working on GGO tumors and estimating the overall GGO coverage of detected nodules in the lung. Current methods utilize manual expert analysis for this important task. We propose to measure quantitatively the opacity property of each pixel in a ground-glass opacity tumor from CT images. Our method results in an opacity map in which each pixel takes opacity value between 0-1. Given a CT image, we propose to accomplish the estimation by constructing a graph Laplacian matrix and solving a linear equations system, with assistance from some manually drawn scribbles for which the opacity values are easy to determine manually. The development of an automated GGO lung tumor detection will greatly improve the efficiency of routine radiological and oncological analysis. Our innovative approach for an objective assessment of GGO tumors will allow the radiologist or thoracic surgeon to evaluate the threedimensional evolution of the tumor and the dimensional changes detected by CT scans taken at different time spans, including changes in growth pattern, maximum areas/orientation of growth, and opacity changes. This proposed study is the first step toward the development of a computerized assessment of GGO tumors and, if successful, will lead to further translational efforts to integrate these techniques into clinical practice. The team brought together to successfully work on this effort is comprised of a thoracic surgeon, who acts as a clinical subject matter expert, and experienced researchers in image enhancement, automated vision and biomedical imaging.