Biomedical Image Processing

该项目使用图像处理技术来分析多种类型的生物、临床和生物医学图像。目前的研究集中在四个一般领域：(1)利用电子显微图像处理的大分子结构生物学；(2)医学成像，包括(A)PET，(B)语音病理中的超声，(C)心脏成像中的超声，(D)CT成像，(E)MRI，(F)EPR成像，(G)癌症研究中的成像，以及(H)与神经功能障碍相关的成像；(3)普通实验室成像；以及(4)生物医学三维重建算法的发展。计算生物科学和工程实验室图像处理研究科与各研究所进行了长期的合作研究工作，涉及使用图像处理技术和先进的计算技术来分析电子显微图像，以确定大分子结构。最近的工作集中在二十面体病毒衣壳结构的三维重建、分析和解释上。计算生物科学和工程实验室图像处理研究科长期致力于为NIH的各种临床和生物医学活动提供计算和工程专业知识。具体地说，PET、超声、CT、MRI、EPR、癌症研究中的成像以及与神经功能障碍相关的成像已通过多种方式得到支持。例如，我们参与了新的动物PET扫描仪技术的开发，使图像分辨率远远超过了以前最先进的扫描仪。为了支持NIH内部计划的科学研究，CIT在开发平台无关的、n维的、通用的、可扩展的图像处理和可视化程序方面取得了重大进展，该程序是用Java编程语言编写的。MIPAV(医学图像处理分析和可视化)应用程序可以对多种形式(如PET、MRI、CT、显微镜等)的医学图像进行定量分析和可视化。它已被用于分析肿瘤的诊断放射学，与NIDCR合作协助进行纵向研究，分析NIMH的MRI图像，并已被NCI用于2D显微样本的分析。我们还为国家心肺和血液研究所(NHLBI)的心脏病学分部开发了名为NIH MAGIC的独特成像软件，这将有助于心脏组织活性研究的发展领域。康复医学部的语言病理学部门也在使用该软件进行3D可视化，目前正在扩展用于其他成像项目。此外，我们还将以前的3-D PET重建算法应用于新一代廉价的小型动物PET扫描仪，该扫描仪由相对排列的像素化闪烁晶体组成。由于这些像素化扫描仪适用于大容量小动物研究，我们正在开发客户端-服务器软件，以促进在CIT的高性能计算集群上进行生产模式重建处理。我们开发并实现了EPR重建的迭代算法，包括乘法算术重建技术(MART)和最小二乘/熵最大化算法。我们目前在EPR方面的工作集中在使这些算法适应非常噪声的成像环境，例如高梯度脉冲EPR。

This project uses image processing techniques to analyze many types of biological, clinical and biomedical images. Current research focuses on four general areas: (1) the structural biology of macromolecules using image processing of electron micrographs; (2) medical imaging, which includes (a) PET, (b) ultrasound in speech pathology, (c) ultrasound in cardiac imaging, (d) CT imaging, (e) MRI, (f) EPR imaging, (g) imaging in cancer research, and (h) imaging related to neural disfunction; (3) general laboratory imaging; and (4) biomedical 3D reconstruction algorithm development. The Image Processing Research Section, Computational Bioscience and Engineering Laboratory has a long term collaborative research effort with the institutes involving the use of image processing techniques and advanced computational techniques to analyze electron micrographs with the goal of determining macromolecular structures. Recent efforts have concentrated on the 3D reconstruction, analysis and interpretation of the structures of icosahedral virus capsids. The Image Processing Research Section, Computational Bioscience and Engineering Laboratory has a long-term commitment to providing computational and engineering expertise to a variety of clinical and biomedical activities at NIH. Specifically, PET, ultrasound, CT, MRI, EPR, imaging in cancer research, and imaging related to neural disfunction have been supported in a number of ways. For example, our participation in the development of new animal PET scanner technology has extended image resolution far beyond what was available from previous state-of-the-art scanners. To support scientific research in the NIH intramural program, CIT has made major progress in the development of a platform-independent, n-dimensional, general-purpose, extensible image processing and visualization program written in the JAVA programming language. The MIPAV (Medical Image Processing Analysis and Visualization) application enables quantitative analysis and visualization of medical images of numerous modalities (i.e. PET, MRI, CT, microscopy, etc.). It has been used to analyze tumors for Diagnostic Radiology, assisted in longitudinal studies in collaboration with NIDCR, analysis of MRI images for NIMH, and has been used by NCI for the analysis of 2D microscopic samples. We have also been instrumental in the development of unique imaging software called NIH magic for the Cardiology Branch of the National Heart Lung and Blood Institute (NHLBI) which will contribute to the developing field of cardiac tissue viability studies. This software is also being used by Speech pathology in the Department of Rehabilitation Medicine to perform 3D visualization and is currently being extended for use in other imaging projects. In addition, we have adapted the our prior 3-D PET reconstruction algorithms for a new generation of inexpensive small animal PET scanners, consisting of opposed arrays of pixelated scintillation crystals. Since these pixelated scanners are suitable for high-volume small animal studies, we are developing client-server software to facilitate production mode reconstruction processing on CIT's high performance computing cluster. We have developed and implemented iterative algorithms for EPR reconstructions, including the multiplicative arithmetic reconstruction technique (MART) and a least-squares/entropy maximization algorithm. Our current work in EPR is focusing on adapting these algorithms to very noisy imaging environments, e.g., high-gradient pulsed EPR.