Continued Development of 4-dimensional Image Warping and Registration Software

摘要：本项目旨在继续开发、测试和评估一种四维（4D）图像翘曲和配准算法，重点是测量大脑结构及其随时间的演变。医学成像的爆炸式增长创造了对定量、高度自动化、准确和强大的医学图像分析软件工具的开发需求。图像配准引起了特别的科学兴趣，因为它对于整合和比较来自个人或群体的数据以及开发反映一组个人内部解剖和功能变化的统计地图集是必要的。此外，它还可以通过对同一个体的连续扫描进行共登记来测量结构和功能的时间演变，并最终使用它们构建4D统计地图集。虽然大量的三维图像配准方法已经被开发和广泛使用，但目前还缺乏可靠的方法来测量细微的纵向变化，即结构和功能随时间的变化。目前，纵向变化通常是通过首先对一系列单个扫描应用标准3D配准方法，然后对它们应用某种回归来测量的。然而，由于三维配准方法的精度有限，理论和实验都知道，估计的纵向变化与实际的变化有很大的不同，更重要的是，当将变化作为两个独立的三维测量值之间的差异进行评估时，各种噪声被放大了。我们开发了一种可变形的配准算法，称为HAMMER，并且是首批将其扩展到全4D可变形配准方法（4D-HAMMER）的先驱之一，用于同时向地图集同步注册串行扫描，并展示了与类似的3D配准相比，在测量串行扫描中形态变化的细微模式方面的显着改进。该方法同时分析个体的所有序列扫描，从而可以评估纵向变化并同时将序列扫描记录到图谱中。这一过程有效地减少了噪声的有害影响，因此在测量纵向变化时更加准确。显然，为4D-HAMMER设计一个易于使用、功能强大的软件包，并将其整合到ITK中，将使大量需要使用这种先进图像分析方法进行纵向研究的最终用户受益。为了提高4D-HAMMER对纵向数据的可变对比度和质量的鲁棒性，并显著提高其速度，还需要进一步的算法开发。为了增加非计算机分析方法专家使用的便利性，并将该软件集成到ITK中，还计划进行重大的软件工程工作。因此，在本项目中，我们将a)进行新的方法开发，在单一框架内集成和优化4D组织分割和配准，以使4D- hammer更加准确和鲁棒，并通过将计算时间从20多个小时减少到几个小时来显着加快4D- hammer（目标1）；b)进行一系列验证实验和对比研究，优化新4D-HAMMER的参数，并将其与之前的方法以及两种广泛使用的替代方法，例如SPM和基于b样条的翘曲算法进行比较，以证明其性能（Aim 2）；c)开发和传播用户友好且文档完备的软件，并将其纳入ITK（目标3-4）。公共卫生相关性：该项目旨在继续开发、测试和评估一种四维（4D）图像扭曲和配准算法，重点是测量大脑结构及其随时间的演变。为了提高四维图像变形算法对纵向数据对比度和质量变化的鲁棒性，并显著提高其速度，还需要进一步的算法开发。重要的软件工程工作也计划增加非计算机分析方法专家的易用性，并将该软件集成到ITK中，以使需要访问这种高级图像分析方法进行纵向研究的大型最终用户社区受益。在建议的方法和软件发展后，我们将向公众提供一个最终的方便用户和文件完备的软件包，并将其纳入资讯科技资讯库。

DESCRIPTION (provided by applicant): Continued Development of 4-dimensional Image Warping and Registration Software Abstract: This project aims to continue the methodological development, testing and evaluation of a 4-dimensional (4D) image warping and registration algorithm, with emphasis on measurement of brain structure and its evolution over time. The explosive growth of medical imaging has created the need for the development of quantitative, highly automated, accurate, and robust software tools for analysis of medical imagery. Image registration has attracted particular scientific interest, since it is necessary for integration and comparison of data from individuals or groups, as well as for the development of statistical atlases that reflect anatomical and functional variability within a group of individuals. Moreover, it can be used to measure the temporal evolution of structure and function by co-registering serial scans from the same individual, and ultimately using them to construct 4D statistical atlases. Although a plethora of 3D image registration methods has been developed and used extensively, there is currently a scarcity of methods for robustly measuring subtle longitudinal change, i.e., change of structure and function with time. Currently, longitudinal change is typically measured by first applying a standard 3D registration method to individual scans in a series and then applying some sort of regression to them. However, the estimated longitudinal change is known both theoretically and experimentally to be significantly different from the actual one, due to the limited accuracy of the 3D registration methods, and more importantly to all sorts of noises that are amplified when evaluating change as a difference between two independent 3D measurements. We have developed a deformable registration algorithm, called HAMMER, and was among the first groups to pioneer its extension to a fully 4D deformable registration method (4D-HAMMER) for co- registering serial scans to an atlas simultaneously, and to demonstrate the significant improvement compared to analogous 3D registration in measuring subtle patterns of morphological change in serial scans. This approach analyzes all serial scans of an individual at the same time, thus it can assess longitudinal change and register serial scans to the atlas simultaneously. Effectively, this process reduces the detrimental effects of noise, and therefore is more accurate in measuring longitudinal changes. Clearly, designing an easy-to-use, robust software package for 4D-HAMMER and incorporating it into the ITK will benefit a large community of end-users that need access to this advanced image analysis method for longitudinal study. To increase the robustness of the 4D-HAMMER to the variable contrast and quality of longitudinal data, as well as to significantly improve its speed, further algorithm development is necessary. To increase ease of use by non-experts in computer analysis methods and to integrate this software into the ITK, significant software engineering efforts are also planned. Therefore, in this project, we will a) carry out novel methodological developments to integrate and optimize the 4D tissue segmentation and registration within a single framework for rendering the 4D-HAMMER more accurate and robust, and to significantly speed up the 4D-HAMMER by reducing the computational time from over 20 hours to several hours (Aim 1); b) perform a series of validation experiments and comparative studies, to optimize parameters in the new 4D-HAMMER and demonstrate its performance by comparing it with previous one, as well as with the two widely used alternative methods, e.g., SPM and B-spline based warping algorithms (Aim 2); c) develop and disseminate user-friendly and well-documented software, and also incorporate it into ITK (Aims 3-4). PUBLIC HEALTH RELEVANCE: This project aims to continue the methodological development, testing and evaluation of a 4-dimensional (4D) image warping and registration algorithm, with emphasis on measurement of brain structure and its evolution over time. To increase the robustness of the 4D image warping algorithm to the variable contrast and quality of longitudinal data, and to significantly improve its speed, further algorithm development is necessary. Significant software engineering efforts are also planned to increase ease of use by non-experts in computer analysis methods, and to integrate this software into the ITK to benefit a large community of end-users that need access to this advanced image analysis method for longitudinal study. With proposed methodological and software developments, a final user-friendly and well- documented software package, with incorporation into the ITK, will be available to the public.