Broad Area Colloquium For AI-Geometry-Graphics-Robotics-Vision

Model-Based Analysis of Medical Images

James S. Duncan, Ph.D.
Professor of Diagnostic Radiology and
Electrical Engineering
Director, Biomedical Engineering Program
Yale University
Monday, March 10, 2002, 4:15PM
TCSeq 201


The development of methods to accurately and reproducibly recover useful quantitative information from medical images is often hampered by uncertainties in handling these data related to: image acquisition parameters, the variability of normal human anatomy and physiology, the presence of disease or other abnormal conditions, and a variety of other factors. This talk will review image analysis strategies that make use of models based on geometrical and physical/biomechanical information to help constrain the range of possible solutions in the presence of such uncertainty. The discussion will be focused by looking primarily at three problem areas in the realms of cardiac function analysis, neuroanatomical structure analysis and brain shift issues in neurosurgery. In the cardiac area, we have developed a set of algorithms that can derive in vivo strain patterns from noninvasive four dimensional (3 spatial dimensions plus time) image datasets, including Magnetic Resonance Imaging (MRI) and ultrasound. This information is then used in conjunction with biomechanical models of the LV to estimate dense strain patterns in the myocardium, which in turn are used to study the physiology of the heart in both normal and abnormal states. In the analysis of neuroanatomical structure, both parametric and level set-based curve evolution strategies are used to segment subcortical and cortical gray matter in the brain. In addition, a variety of shape properties of these regions are used along with volume estimates to study patient populations with a variety of neurological disorders. Finally, we note that during epilepsy neurosurgery, a multi-modality (MRI, MR spectroscopy, EEG, SPECT) assortment of images and signals are acquired to give the neurosurgeon a roadmap of the patient's brain in order to guide surgical/interventional procedures aimed at eliminating severe seizure activity. We have used a variety of mathematical reasoning strategies to integrate or fuse this preoperative information. In addition, we are currently working on developing a system that makes use of stereo cameras mounted in the operating room, along with a linear elastic biomechanical model of the brain, to account for the brain shift that occurs during the surgery. Accurately accounting for this will greatly improve the utility of our current image-guided surgery navigation system. The presentation will include a description of the problem areas and visual examples of the image datasets being used, an overview of the mathematical techniques involved and a presentation of results obtained when analyzing actual patient image data using these methods. Emphasis will be placed on how image-derived information and appropriate modeling can be used together to address the image analysis and processing problems noted above.

About the Speaker

James S. Duncan received the BSEE degree from Lafayette College, Easton, PA in 1973, the MS degree in Engineering from UCLA in 1975 and the Ph.D. degree in Electrical Engineering from the University of Southern California, in 1982. In 1973, he joined the staff of Hughes Aircraft Company, Electro-Optical and Data Systems Group, and participated in research and development projects related to signal and image processing for forward looking infrared (FLIR) imaging systems until 1983. During this time, he held Hughes' Masters, Engineer and Doctoral Fellowships. In 1983, he joined the faculty of Yale University, New Haven, CT., where he currently is a Professor of Diagnostic Radiology and Electrical Engineering; is the Vice-Chair of Bioimaging Sciences Research and the Director of the Image Processing and Analysis Group within Diagnostic Radiology; and is the Chair of Yale's Program in Biomedical Engineering. His research and teaching efforts have been in the areas of image processing, computer vision and medical imaging. His current specific research interests include the segmentation of deformable objects from both 2D and 3D data, the tracking of nonrigid object motion from 2D and 3D data, the use of geometrical and phyiscal models to constrain the recovery of information from images and the integration of processing modules in vision systems, all with a special interest in using these approaches for medical image analysis. Dr. Duncan is a member of Eta Kappa Nu and Sigma Xi, is on the editorial board of the Journal of Mathematical Imaging and Vision, and is currently an Associate Editor for the IEEE Transactions on Medical Imaging and a co- Editor of the journal Medical Image Analysis. He was a Fulbright Research Scholar at the Universities of Amsterdam and Utrecht in the Netherlands during part of the 1993-94 academic year. In June, 1997, he chaired the international conference on Information Processing in Medical Imaging (IPMI), held in Poultney, Vermont. In 1999, he was elected as Fellow of the American Institute for Medical and Biological Engineering (AIMBE) and in 2000 he was elected to be a Fellow of the IEEE for contributions in medical image analysis and computer vision.

Back to the Colloquium Page