Estimation of tissue deformation in medical images

在过去的十年里，我们看到计算机在医学领域的应用有了显著的增长。医学图像分析技术的广泛使用帮助临床医生超越了传统医学的许多局限性。我的研究重点在于同样的范式：开发新的医学图像分析技术，是可靠的临床应用。更具体地说，我的研究项目涉及超声（US）弹性成像和超声图像与磁共振（MR）和计算机断层扫描（CT）的融合/配准的新计算技术的发展。******US是一种易于使用，价格低廉，实时和安全的成像方式。这种相对年轻的医学成像方式也是最常用的一种，并且正在迅速发展。事实上，许多新的研究项目最近已经转化为商业产品。这项研究计划有两个重点，可以进一步提高美国的能力。首先，我们将开发新的图像处理技术，准确地跟踪组织变形与US创建新的图像，显示组织的机械性能。这是重要的临床影响，因为病理变化的组织，如癌症肿瘤，是高度相关的组织力学性质。其次，我们将开发鲁棒和准确的技术，自动变形注册US到MR和CT。图像配准涉及对从同一组织获得的两幅图像进行对齐，以便对它们进行比较或组合。磁共振和CT的US注册在诊断和治疗/干预方面有许多应用，是一个活跃的研究领域。在这两个推力中，我们的方法是将问题转换为包含图像相似性和先验信息的成本函数的优化。为了实现广泛的临床应用，弹性成像和注册的一个关键要求是实时性能。提出的研究计划将通过开发新的计算高效的图像分析技术来实现这一目标。******拟议的跨学科研究计划将在康考迪亚大学电气与计算机工程系和PERFORM中心的一个令人兴奋的环境中培养两名博士和四名MASc学生。

We have seen a significant increase in the use of computers in medicine in the past decade. Extensive use of medical image analysis techniques has helped clinicians transcend many limitations of conventional medicine. The focus of my research lies in the same paradigm: to develop novel medical image analysis techniques that are reliable for clinical use. More specifically, my research program involves the development of novel computing techniques for ultrasound (US) elastography and fusion/registration of US images to magnetic resonance (MR) and computed tomography (CT).******US is an easy-to-use, inexpensive, real-time and safe imaging modality. This relatively young medical imaging modality is also one of the most frequently used, and is rapidly growing. In fact, numerous new research projects have recently been translated into commercial products. This research proposal has two thrusts that further improve the capabilities of US. First, we will develop novel image processing techniques that accurately track tissue deformation with US to create new images that show tissue mechanical properties. This is of significant clinical impact since pathologic changes in tissue, such as cancer tumours, are highly correlated with tissue mechanical properties. Second, we will develop robust and accurate techniques for automatic deformable registration of US to MR and CT. Image registration involve aligning two images that are obtained from the same tissue, so that they can be compared or combined. Registration of US with MR and CT has numerous applications in both diagnosis and therapy/interventions, and is an active field of research. Our approach in both thrusts is to cast the problem as optimization of a cost function that incorporates image similarity and prior information. To achieve widespread clinical use, a critical requirement in both elastography and registration is real-time performance. The proposed research program will achieve this goal by developing novel computationally efficient image analysis techniques.******The proposed interdisciplinary research program will train two PhD and four MASc students in an exciting environment at the Department of Electrical and Computer Engineering and the PERFORM Centre at Concordia University.