BIOMEDICAL IMAGE PROCESSING

点击翻译按钮获取中文摘要

This project uses image processing techniques to analyze many types of biomedical images. Current research includes the structural biology of macromolecules, analysis of electron micrographs, cataract analysis using images of the eye, functional PET scans to analyze the relationship between brain anatomy and brain function, MRI and functional MRI scans, and ultrasound images of the heart to analyze blood flow. To answer important questions in structural biology, it is necessary to obtain high resolution 2- and 3-D structural information about biological macromolecules. Biological specimens can be visualized in the electron microscope using cryo-electron microscopy, a newer technique, which attempt to preserve "native" structure by surrounding the specimen with a thin laye of ice. Of particular interest is the understanding of viral structures. At present we are continuing our efforts to investigate the structure of a large animal virus, human herpes simplex virus (type 1). Most recent work i this field has centered on how the herpesvirus capsid is initially formed, and time based studies are being used to show the maturation and changes in the capsid structure. Very high resolution 3D reconstructions of the bovine papillomavirus (bpv) capsid are being studied. This analysis, currently at 9A, leads the way to high resolution among all laboratories in the world. New computational, as well as biochemical and microscopic techniques, are being developed to achieve this phenomenal resolution. In the field of medical imaging, with NEI we have been developing a series of images of cataracts which could be utilized as standards for the physician to subjectively grade cataracts. These standards are based upon numerical quantitation obtained from the Zeiss Scheimpflug system which use software developed by us for on-line quantitation. We are also beginning to develop a slit lamp which can be used by many sites for quantitative gradin purposes. We have developed unique medical application software for use by the Cardiology Branch of the National Heart, Lung and Blood Institute (NHLBI) which should contribute to the developing field of cardiac tissue viability studies. The software should prove to be the most accurate system yet designed to estimate myocardial tissue viability as a result of contras agents reaching the coronary microcirculation. The computer analysis of functional brain magnetic resonance imaging (MRI) images facilitates understanding of problems ranging from developmental learning disorders to language disabilities arising from stroke. We have applied Principal Component Analysis (PCA) multivariate technique to compress the fMRI data sets. The advantages of PCA are twofold, that principle components are orthogonal to each other and that any subsequent analysis of some other variable may be more easily performed. In addition, filtering programs were created for analyzing PET image brain tissue voxels for the Aging Institute using Gaussian convolutions and median filters. These programs used gray matter, white matter, and CSF information obtained from stereotactically aligned and segmented MRI images.