Framework for multi-feature registration of medical images

医学图像多特征配准框架

• 批准号: RGPIN-2015-03633
• 负责人: Collins, DLouis
• 金额: $ 5.32万
• 依托单位: McGill University
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2016
• 资助国家: 加拿大
• 起止时间: 2016-01-01 至 2017-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=612468
• 关键词: Framework; multi; feature; registration; medical

Framework for multi-feature registration of medical images Millions of dollars of government and industrial support have been used to collect medical image data from the living brains of tens of thousands of subjects with the goal to advance our understanding of the human brain in health and in disease. Recent advances in computational technology and computer science made it possible to analyze such enormous quantities of data. Examples brain analysis can include the study of longitudinal growth trajectories in normal development or in abnormal development; the study of ageing to help understand disease progression; and the identification of imaging biomarkers for diagnostic purposes. Most analyses of brain data relies on a registration step where all brain are aligned into a standard coordinate system, so that an x,y,z coordinate addresses the same anatomical location in all brains. However, analysis of the terabytes of data representing thousands of subjects can be time consuming even when implemented on a supercomputer or cluster. This constraint limits the interactive analysis of such data. New tools are needed to align these datasets so that useful information can be extracted from the terabytes of data collected. The long-term objectives the proposed research program are to build novel image processing techniques, based on probabilistic/mathematical frameworks to improve the registration tools used in brain analysis. In our short term objectives, we propose to develop tools to enable learning the statistical variability of healthy neuroanatomy across large populations of subjects and model longitudinal change in specific sub-groups of the population. These models will be used to enhance and improve the robustness and accuracy of registration, tissue classification and structure segmentation techniques when applied to large databases of brain MRI. These tools will lead to improved neuroscientific research that may in turn lead to concrete improvements in medical diagnosis and patient care in the form of novel image-based biomarkers of disease.

医学图像多特征配准框架 数百万美元的政府和工业支持已被用于从数万名受试者的活体大脑中收集医学图像数据，目的是促进我们对健康和疾病中人类大脑的理解。计算技术和计算机科学的最新进展使得分析如此大量的数据成为可能。大脑分析的例子可以包括研究正常发育或异常发育中的纵向生长轨迹;研究衰老以帮助了解疾病进展;以及识别用于诊断目的的成像生物标志物。大脑数据的大多数分析都依赖于一个配准步骤，在这个步骤中，所有的大脑都被对齐到一个标准坐标系中，这样x，y，z坐标就可以在所有大脑中找到相同的解剖位置。然而，即使在超级计算机或集群上实现，对代表数千个主题的TB级数据的分析也可能是耗时的。这一制约因素限制了对此类数据的交互式分析。需要新的工具来调整这些数据集，以便从收集的TB数据中提取有用的信息。 拟议的研究计划的长期目标是建立新的图像处理技术，基于概率/数学框架，以改善大脑分析中使用的配准工具。在我们的短期目标中，我们建议开发工具来学习大量受试者群体中健康神经解剖学的统计变异性，并对特定人群亚组的纵向变化进行建模。这些模型将用于增强和提高注册，组织分类和结构分割技术的鲁棒性和准确性时，适用于大型数据库的脑MRI。 这些工具将导致神经科学研究的改进，从而可能以新的基于图像的疾病生物标志物的形式导致医疗诊断和患者护理的具体改进。