FiberQuant: A Tool for Rapid Quantitative Analysis of White Matter Fiber Tracts

描述（由申请人提供）：本SBIR项目将开发一个医疗器械软件系统“FiberQuant”，用于根据扩散张量成像（DTI）磁共振图像提供的数据自动识别和分析白质纤维束。第一阶段的工作将首先通过实施方法并验证其在对照人群中的应用来证明可行性，但最终的技术将对几种神经和精神疾病有用。如果要使这些方法有效地自动化并过渡到适合常规临床使用的医疗设备，就必须克服重大的技术挑战。为了应对这些挑战，在第一阶段，我们将实施DTI数据预处理方法，包括涡流失真、B0失真和运动校正。基于CorTechs实验室现有的专利和商业化的基于地图集的分割方法，我们还将创建并应用DTI光纤地图集来驱动自动轨迹识别算法。自动衍生纤维将通过与人工衍生纤维的比较来验证。最后，我们将开发和应用无偏的方法来量化自动识别纤维束中的白质特性。由此产生的软件工具有望具有广泛的科学和临床用途。CorTechs实验室之前已经展示了开发智能图像分析方法的能力，该方法可以从MRI测量中获得大脑解剖学的定量测量，并将这种方法转化为商业的，fda批准的医疗设备。假设从第一阶段成功过渡，在第二阶段，我们将通过开发扩散衍生测量的规范数据库来扩展这些技术，使医生能够可视化纤维束的部分，这些部分显示出与对照人群均值的显著偏差。将对各种患者组进行验证研究，以确定这些工具是否可以提供识别异常白质组织的方法。经过适当验证，这些工具将具有诊断用途以及手术前计划的实用性。我们最终计划在临床测试站点安装中评估这些工具，并获得FDA许可，使其成为适合常规临床使用的医疗器械软件。公共卫生相关性：基于扩散张量成像磁共振数据，越来越需要有效和高效的计算方法来自动识别和量化人脑中单个白质纤维束，以及特定纤维束内的潜在异常。这些工具可用于医学诊断和手术前规划，以及生物医学和制药研究。在这两种应用中，单个白质纤维束的识别是至关重要的，以便定位功能相关结构的病理变化，以及提供兴趣区域，从而提高相对于体素比较的统计能力。这个项目将提供的工具有很大的潜在市场。1

DESCRIPTION (provided by applicant): This SBIR project will develop a medical device software system, "FiberQuant", for the automatic identification and analysis of white matter fiber tracts based on data provided by diffusion tensor imaging (DTI) magnetic resonance images. The Phase I effort will first demonstrate feasibility by implementing methods and validating their application in a control population, but the resulting technology will be useful with regard to several neurological and psychiatric disorders. Major technical challenges must be overcome if such methods are to be effectively automated and transitioned into a medical device suitable for routine clinical use. To address these challenges, in Phase I we will implement methods for preprocessing DTI data that will include correction for eddy current distortion, B0 distortion, and motion. Building on CorTechs Labs' existing patented and commercially marketed atlas-based segmentation methods we will also create and apply a DTI fiber atlas to drive an automated track identification algorithm. Automatically derived fibers will be validated by comparison to manually derived fibers. Finally we will develop and apply unbiased methods for quantifying white matter properties in the automatically identified fiber tracts. The resulting software tools promise to have widespread scientific and clinical utility. CorTechs Labs has previously demonstrated the capability of developing intelligent image analysis methods for deriving quantitative measures of brain anatomy from MRI measurements, and in transitioning such methods into commercial, FDA-cleared medical devices. Assuming successful transition from Phase I, in Phase II we will extend these techniques by developing a normative database of diffusion-derived measures, to allow physicians to visualize the parts of fiber tracts that show significant deviation from the mean of a control population. Validation studies with a variety of patient groups will be performed to determine whether such tools would provide a means to identify abnormal white matter tissue. Properly validated, these tools will have diagnostic use as well as utility for presurgical planning. We ultimately plan to evaluate these tools in clinical beta site installations, and obtain FDA clearance for them as medical device software suitable for routine clinical use. 1 PUBLIC HEALTH RELEVANCE: There is a growing need for effective and efficient computational methods for automatically identifying and quantifying individual white matter fiber tracts in the human brain -- and potential abnormalities within given fiber tracts -- based on diffusion tensor imaging magnetic resonance data. Such tools could be useful for medical diagnosis and presurgical planning, as well as biomedical and pharmaceutical research. In both applications, the identification of individual white matter fiber tracts is of central importance, in order to locate the pathological changes in functionally relevant structures, as well as to provide regions-of-interest that improve statistical power relative to voxel- wise comparisons. There is a large potential market for the tools this project will provide. 1