This project aims to improve the accuracy of breast cancer diagnosis and increase the chance of early detection by developing an advanced image processing technique for a digital mammography system. To achieve these aims, an iterative, non-linear Bayesian image estimation technique will be implemented for scatter reduction in digital mammography. The major goals of this study are four-fold:

1. Develop and implement a Bayesian image processing technique to reduce x-ray scatter in digital mammography.
2. Compare the effect of this technique on improving image contrast-to-noise ratios (CNR) over that obtained with a grid for several types of phantom images.
3. Investigate this technique as a means to reduce patient dose by replacing the anti-scatter grid.
4. Visually inspect patient images that have been processed by this technique to explore any potential clinically relevant problems.

Preliminary results suggest that Bayesian image processing can be used with digital mammographic images to effectively reduce x-ray scatter and concurrently improve image CNR, without adversely affecting image resolution.

The likelihood of effective treatment of breast cancer is improved by early detection of cancerous tumors. The proposed technique will increase the CNR of masses, while maintaining image resolution. This CNR improvement should increase the probability of detection, especially early detection, of masses. The Bayesian technique will sharpen the objects in the mammographic image and improve the visualization of normal breast tissue, thereby decreasing the number of false positive readings. The improved CNR of the proposed technique will also lead to a method of reducing patient radiation dose. Furthermore, with the coming advent of digital mammography, concern exists about potential aliasing artifacts caused by grid line patterns that may cause misdiagnosis. The technique proposed here will be used to remove the grid from the standard mammographic imaging protocol. Removal of the grid will allow for the exposure to be reduced by the Bucky factor, thus reducing patient dose.

Ki. H. Chon, Ph.D.
Assistant Professor, Department of Molecular Pharmacology
Brown University (97-0367)
Providence, RI

Nonlinear Analysis of Cardiovascular Autonomic Nervous System: Detection and Comparison of Sympathovagal Balance in Healthy and Diseased Conditions

Normal heart rate is determined by many interacting systems, including the nervous system, baro- and chemoreceptors, local feedback loops in the heart, and several hormonal systems. These systems work on different time scales, and the complexity of this regulation is reflected in the apparent random fluctuations in heart rate, a phenomenon termed heart rate variability (HRV). There is considerable interest in these fluctuations because their simple statistical measures such as the standard deviation of the interbeat intervals have been shown to be some of the strongest independent predictors of mortality after myocardial infarction.

Clinically-reliable assessment of interactions between sympathetic and vagal nerve activities requires accurate detection techniques. This project will concentrate on using nonlinear dynamic analyses, including nonlinear determinism theory and nonlinear systems analysis, to quantify these interactions. Changes in the ratio of nonlinear low frequency components (LF) to high frequency components (HF) between normal and diseased subjects can thus be obtained. An increased LF/HF ratio of the linear power spectral density of the heart rate variability (HRV) signal has been associated with patients suffering from acute myocardial infarction.

The three specific aims are to:

1. Quantify the differences in interactions between normal and diseased conditions such as myocardial infarction and chronic heart failure (CHF).
2. Determine if the nonlinear LF/HF ratio provides additional information about the sympathetic-vagal balance.
3. Validate the nonlinear analysis techniques using data to be collected invasively from healthy swine.

By using accurate nonlinear systems analysis algorithms, the proposed approach has the opportunity to make a significant contribution to the understanding of the role of autonomic function imbalance in diseased myocardial conditions.

Amy C. Courtney, Ph.D.
Staff, Department of Biomedical Engineering
The Cleveland Clinic Foundation (97-0368)
Cleveland, OH

Distinguishing Effects of Aging and Disease on Bone Quality

The overall goal of this project is to develop a suite of complementary mechanical, structural, biological and quantitative histological measures to distinguish effects of aging and disease on bone quality. By using these complementary tests, it is hoped to also identify whether changes observed are due to material or structural changes and whether these can be correlated with cellular activity. This goal will be achieved in the context of a specific and important problem, that is, whether the effects of diabetes on bone may be likened to those of (perhaps accelerated) aging or whether they are unique. It is hypothesized that diabetes-related bone fragility is fundamentally different from that which occurs in aging.

The PI will investigate these issues for bones of the feet from cynomolgus monkeys. Foot bones were chosen because individuals with diabetes are particularly susceptible to a spectrum of osteopathies and arthropathies in the foot and ankle, and one-quarter of hospitalizations due to diabetes are for foot problems.

The first aim is to distinguish effects of aging and diabetes on the mechanical properties and susceptibility to microdamage of bone. Standard three-point bending tests will be performed on cortical bone specimens, and mechanical tests will be performed on trabecular bone specimens which (i) minimize artifacts and (ii) correlate with the development of microdamage in trabecular bone.

The second aim is to distinguish effects of aging and diabetes on bone density, geometry, and architecture. While these parameters are coupled in normal aging, the literature suggests that diseases such as diabetes may uncouple them. The measurements will be made using (i) computer image processing techniques based on both spatial and frequency analysis of bone architecture, and (ii) bone densitometry using dual energy x-ray absorptiometry (DXA).

The third aim is to distinguish effects of aging and diabetes on bone metabolism as assessed by (i) serum bone biomarkers, including alkaline phosphatase and osteocalcin, and (ii) the number of osteoclasts and osteoblasts present in tissue sections.

Xiaoyan Deng, Ph.D.
Independent Investigator, Department of Surgery
Laval University (97-0284)
Quebec, Canada

Flow-Dependent Luminal Surface Concentration of Atherogenic Lipoproteins: Role of Hemodynamic Factors in Atherosclerosis

The proposed study is based on the hypothesis that the atherogenic lipoprotein concentration at the blood/vessel wall interface may differ from the bulk concentration, and has an important role in atherogenesis because the blood vessel wall is directly exposed to the luminal surface atherogenic lipoprotein concentration. This concentration may vary according to its location within the circulation, being much higher in regions of disturbed blood flow with low wall shear rates. The high luminal surface lipid concentration in these flow regions can increase the driving potential for the infiltration of the atherogenic lipids into the blood vessel wall, leading to atherosclerosis.

The specific aims are to study:

1. The effect of local blood flow patterns on the spatial distribution of atherogenic lipids at the blood/vessel wall boundary (the luminal surface of the artery).
2. The possible connection between the lipid concentration at the luminal surface and the localization of atherosclerosis.

The experiments will be carried out in 2-dimensional flow chambers that consist of a solid upper wall with a glass corridor and a bottom wall with a semi-permeable section. The semi-permeable section comes from a freshly excised canine aorta. Flow will be steady or pulsatile. The concentration of low-density lipoproteins (LDL) will be measured by fluorescence intensity with a confocal microscope.

By better understanding lipid infiltration and accumulation within arterial prostheses, the PI may find new ways to significantly improve the performance of arterial prostheses and arterial grafting.

Rinat Esenaliev, Ph.D.
Faculty Fellow, Department of Electrical and Computer Engineering
Rice University (97-0370)
Houston, TX

Laser Optoacoustic Imaging for Breast Cancer Detection

Laser optoacoustic imaging is a novel technique recently proposed for cancer diagnostics. Laser optoacoustic imaging utilizes generation of ultrasound by pulsed laser light in biological tissues. Both amplitude and temporal characteristics of the laser-induced ultrasonic (acoustic) waves are dependent on the optical absorption and structure of irradiated tissues. Enhanced light absorption in malignant tumors compared with normal tissues produces ultrasonic sources confined to the tumor. Since the difference in optical absorption is substantially greater than the differences in acoustic properties and x-ray absorption, optoacoustic technique may yield images with substantially better contrast.

One of the most important and promising application for this technology is the diagnosis of small breast tumors with diameters less than 5 mm. The PI proposes to construct a prototype laser optoacoustic system and test it on phantoms and breast tissues. The proposed research will be implemented in three steps:

1. Construct a prototype laser optoacoustic imaging system consisting of an IR laser, a fiber-optic system for laser light delivery, a sensitive acoustic transducer, and a computer for data acquisition, signal processing, and image reconstruction.
2. Test f the system on phantoms simulating breast with small tumors. Absorption coefficient, dimension, shape, and position of the phantom tumors will be varied to study the capability of the system to image them. Experiments with gelatin phantoms and real biological tissues in vitro will be performed.
3. Modify the system in preparation for first clinical trials.

The proposed research will include the comparison of images obtained by the laser optoacoustic system with x-ray, ultrasonic, and MRI images.

Laser optoacoustic tomography promises to provide substantially better image contrast in comparison with x-ray radiography and ultrasonography from radiologically dense and acoustically homogenous phantoms simulating human breast with tumors.

Kevin D. Gillis, D.Sc.
Assistant Professor, Dalton Cardiovascular Research Center
University of Missouri-Columbia (97-0015)
Columbia, MO

Techniques for Membrane Capacitance Measurements in the Presence of Nonlinear Conductances

Electrical measurements of cell membrane capacitance have been used as an exquisitely sensitive, high-time-resolution assay of secretion that is applicable to individual cells. Changes in cell capacitance accompanies the secretion of hormones or neurotransmitter because secretion results in small changes in the surface area of the cell membrane. Use of membrane capacitance measurements has allowed biologists to address previously unresolvable questions about how secretion is triggered, how it is regulated, and the ordering of steps that are involved in secretion.

The general objective of the project is to develop models, algorithms, and experimental protocols for estimation of membrane capacitance as an assay of exocytosis.

The specific aims are:

1. Model the "non-ideal" membrane electrical properties of bovine adrenal chromaffin cells. Non-ideal phenomena include a) time-varying membrane conductance, b) voltage-dependent membrane conductance, c) time and voltage-dependent mobilization of charges with underlie the gating of voltage-dependent ionic channels and d) deviation from the single compartment model of the cell membrane.
2. Devise pulse protocols which minimize the number of "extra" model parameters necessary to describe the cell membrane.
3. Develop techniques for estimating membrane capacitance and other model parameters for the extended membrane models.
4. Verify the new techniques in biological experiments. Results from biological experiments will likely lead to revisions in the simplifying assumptions made in Aim 2 and the techniques developed in Aim 3.

The proposed project aims to provide techniques to allow the time course of secretion during a depolarizing stimulus to be directly measured. This should greatly aid the study of the kinetics of secretion, which is currently a topic of great interest in the biological community.

Robert Guldberg, Ph.D.
Department of Mechanical Engineering
Georgia Institute of Technology (97-0306)
Atlanta, GA

The Effects of Controlled in vivo Mechanical Loading on Bone Graft Repair

The adaptation of bone mass and structural organization to mechanical stimuli has been an important focus of research effort for over a century. However, our ability to clinically exploit the adaptive potential of bone has been impeded by difficulties in assessing cell-mediated responses and the associated local mechanical stimuli from in vivo studies. The hydraulic bone chamber (HBC) model was specifically designed to study the mechanical influences on bone repair and regeneration in vivo. The titanium HBC implant resembles a hollow screw with transverse holes for bone tissue infiltration and combines the physiological relevance of an in vivo system with several of the well-controlled features of in vitro systems.

This system will be used to achieve the following specific aims:

1. To quantify the early effects of intermittent mechanical loading on bone matrix, growth factor, and integrin mRNA expression using quantitative reverse transcription polymerase chain reactions (RT-PCR) and in situ hybridization.
2. To quantify the intermediate effects of intermittent mechanical loading on type I procollagen expression using immunohistochemistry and grid-based stereology.
3. To quantify the later effects of intermittent mechanical loading on new bone matrix formation and organization using quantitative histomorphometry.
4. To correlate spatial distributions of local mechanically-induced responses with spatial distributions of trabecular tissue strain predicted by digital image-based finite element analysis.

A comprehensive and quantitative understanding of mechanical adaptation during bone graft repair may suggest alternative treatment methodologies using traditional bone defect materials and would provide a basis for the design and evaluation of novel bone regenerative constructs.

X. Edward Guo, Ph.D.
Assistant Professor, Department of Mechanical Engineering
Columbia University in the City of New York (97-0086)
New York, NY

Quantification of in vivo Cellular Adaptation of Trabecular Bone by Mechanical Stimulation

The long-term goal of this research program is to systematically quantify the relation between mechanical stimulation and adaptation of bone and to ultimately develop a mathematical law which relates the local stress/strain parameters to cellular/molecular responses of bone tissue. In this project, the temporal and dose-dependent cellular, adaptational responses of trabecular bone to mechanical loading will be quantified utilizing a rat tail vertebra model. A unique feature of this proposal is that the local stress/strain environment in the bone tissue can be fully characterized using micro computed tomography system and newly established three-dimensional image-based finite element modeling techniques. By combining the advantages offered by these state-of-the-art techniques, it is possible for the first time to begin to derive a quantitative relation between local cellular response and local mechanical environment.

The specific aims are:

1. To determine the temporal and spatial patterns of bone cell activities in response to mechanical stimulation.
2. To determine the in vivo relationship between bone cell activities and the mechanical loading magnitude.
3. To determine the in vivo relationship between bone cell activities and local stress/strain environments predicted by computational and microstructural models.
4. To determine the temporal evolution of trabecular morphology and stress/strain distribution in adaptive trabecular bone structure by mechanical stimulation.

Trabecular bone adaptation plays a significant role in the etiology of many metabolic bone diseases such as osteoporosis, osteopetrosis, bone loss in microgravity and the long-term success or failure of porous implants in total joint arthroplasty. Therefore, quantifying the adaptation of trabecular bone to mechanical stimulation may significantly influence the approach to many clinical conditions.

Zhenyu Guo, Ph.D.
Assistant Professor, Department of Electrical Engineering and Computer Science
The George Washington University (97-0344)
Washington, DC

Three-Dimensional Power Doppler Angiography of Lower-Extremity Occlusive Disease

Power Doppler Angiography (PDA) is an ultrasound imaging modality that can produce angiography like images of the flow lumen. However, this new technique has mainly been used to locate vessels and map blood perfusions instead of being used as an angiographic tool to quantify arterial stenoses. The PI has used 3D PDA for quantifying single and multiple stenoses using flow phantoms. According to his results, 3D PDA is a promising technique to replace contrast angiography. However, many fundamental problems related to clinical relevant conditions must be solved before this technique can be applied to clinical practice.

The general objective of the present proposal is to develop 3D PDA into a routine clinical angiographic tool as an alternative method of contrast angiography. The specific aims of this project are:

1. To investigate the accuracy in quantifying stenoses of lower-extremity arteries using 3D PDA.
2. To investigate how flow recirculation affects the quantification of the length of the stenosis.
3. To investigate the influences of flow turbulence and red blood cell aggregation on the 3 D power Doppler angiograms of stenotic vessels and possible solutions.
4. To upgrade the system to be able to perform flow gating image acquisition and be able to image with an extended field-of-view for clinical applications.
5. To perform clinical test on 3D PDA, compare results with those from color Doppler duplex ultrasound and contrast angiography.

This project will result in a clinically relevant 3D PDA system for quantifying lower-extremity arterial stenoses and revealing run-off vessels. Successful completion of this project will produce a great impact on reducing the health care cost on cardiovascular disease as this technique can be expanded to diagnose stenoses of other arterial beds, to monitor bypass graft patency, and to diagnose deep vein thrombosis. Preliminary results and the PI's experience on 3D power Doppler imaging ensures that the objectives described in this proposal are achievable.

Blaise Frederick, Ph.D.
Assistant Biophysicist, Department of Brain Imaging
McLean Hospital (97-0279)
Belmont, MA

Development of a Phased Array Echoplanar Magnetic Resonance System for Functional MRI in Neuropsychiatry

Functional magnetic resonance imaging (fMRI) has become one of the most important new imaging tools in clinical neuroscience, due to its ability to noninvasively measure alterations in cerebral blood flow and neural activation in response to external stimuli or cognitive tasks. The image intensity changes in fMRI caused by neural activation and cerebral blood flow changes are small--on the order of 2% to 5% at most in the case of neural activation. As a result, optimizing the signal to noise ratio (SNR) of fMRI experiments is critical to their success. The PI will design and construct a phased array receiver system for echoplanar magnetic resonance imaging and hopes to enhance the SNR by 30% to 200%. This system simultaneously records magnetic resonance signals from many individual RF receiver coils, each designed to optimize the SNR in a particular region.

The four specific aims are:

1. The design and construction of a phased array echoplanar receiver system.
2. The design and construction of phased array coils optimized for functional imaging experiments.
3. Testing in both phantoms and volunteer subjects to demonstrate the enhanced sensitivity of the system relative to conventional fMRI.
4. The use of the enhanced fMRI system for the evaluation of cerebral perfusion and photic activation experiments in patients with mild to moderate Alzheimer's disease, patients with non-Alzheimer's dementia, and a group of age and gender matched comparison subjects.

The enhanced exam using fMRI yields much the same information as a SPECT exam without the use of radionuclides or the inconvenience of a separate exam. The enhanced spatial resolution may enhance the specificity of this exam for distinguishing Alzheimer's dementia from other conditions.

Assen Kirov, Ph.D.
Instructor, Department of Radiology
Washington University (97-0336)
St. Louis, MO

Three-Dimensional Scintillation Dosimetry Using Tomographic Reconstruction

Brachytherapy is a cancer treatment method whereby a radioactive source is inserted directly into a localized tumor. In conventional brachytherapy the radioactive sources are sufficiently small so that the use of a point source approximation provides satisfying accuracy in calculating the three-dimensional (3D) dose distribution around the source. The use of treatment applicators such as radioactive stents and balloon catheters filled with radioactive fluid with complex geometry requires experimental techniques for determining the three-dimensional dose distribution around them. This need has increased due to the recent development of applicators. Dynamic external beam therapy also lacks an accurate, efficient, and cost-efficient method for two- or three-dimensional dosimetry. The available dosimetric techniques are only able to provide estimate of the dose at a point or a plane.

The PI proposes an investigation to establish the feasibility, build and optimize a new method capable to measure simultaneously the energy deposited by ionizing radiation in all points of a volume filled with material with dosimetric properties close to these of water. The method will be based on the scintillation process and the work on the project will include: (i) investigation and optimization of the properties of radiation dose sensitive media; (ii) design and construction of a data acquisition system; (iii) modification of certain iterative reconstruction algorithms; and (iv) develop calibration procedures and validate dose measurements.

Joseph Y. Lo, Ph.D.
Assistant Research Professor, Department of Radiology
Duke University Medical Center (97-0322)
Durham, NC

Artificial Neural Networks for Predicting Breast Cancer Invasion

The main objective of this study is to develop a computer-aided diagnosis system to predict breast cancer malignancy and invasion among nonpalpable, mammographically suspect lesions. This CADx system will consist of artificial neural network (ANN) decision models which will process findings from the patient's history and mammographic findings extracted by expert mammographers. The goal is to predict whether mammographic lesions are benign, in situ cancer, or invasive cancer. This CADx system may assist in surgical planning for patients with breast lesions, and can reduce the cost and morbidity of unnecessary surgical biopsies.

The specific aims of this study are as follows:

1. Prospectively acquire a large patient findings database for robust training and testing of ANNs.
2. Optimize an ANN to distinguish between invasive and in situ carcinomas by using malignant cases only.
3. Investigate unified approaches to predicting both malignancy and invasion for all patients.
4. Search for optimal subsets of findings that simplify data collection and computation while maintaining performance.
5. Perform cost-effectiveness analysis to determine the optimal trade-off between sensitivity and specificity for the CADx system.

Currently benign lesions and invasive cancers together comprise up to 90% of cases undergoing surgical biopsy. Compared to excisional or stereotaxic biopsy, the proposed CADx system may provide similarly accurate diagnosis for certain specific groups of patients while being completely noninvasive. Furthermore, this CADx system may play a complimentary role by identifying patients most suitable for stereotaxic biopsy, thus significantly reducing cost and morbidity compared to open excisional biopsy.

Howard W. T. Matthew, Ph.D.
Assistant Professor, Departments of Chemical Engineering and Materials Science
Wayne State University (97-0361)
Detroit, MI

In vitro Analysis and Control of Colonization and Remodeling in a Tissue Engineered Vascular Prosthesis

Synthetic vascular grafts less than 6 mm inner diameter suffer high failure rates for primarily three reasons. First, poor blood compatibility leads to blood clot formation and obstruction. Second, excessive proliferation of vascular smooth muscle cells (SMC) may lead directly to vessel obstruction, or indirectly to blood clot formation and subsequent blockage. Finally, poor matching of mechanical properties between grafts and natural vessels can lead to leakage, aneurysm, or structural failure. These problems have limited the usefulness of vascular prostheses for small blood vessels and ultimately result in less than ideal treatments or multiple surgical procedures for a single problem.

The proposed studies are founded on the hypothesis that superior results can be obtained by constructing composite vascular grafts from resorbable materials which are overtly anti-thrombogenic and which specifically moderate the proliferation of vascular smooth muscle while allowing a rapid endothelialization of the luminal surface. It is believed that composites synthesized from complexes of the ionic polysaccharide chitosan and the glycosaminoglycans (e.g. heparin) hold considerable promise in this area. The specific aims of this project are:

1. To optimize methods for controlling the microstructure and three-dimensional microscale composition of vascular prostheses synthesized from bioactive polysaccharide materials.
2. To design and characterize a perfusion culture system for monitoring and evaluation of vascular graft colonization and remodeling in vitro.
3. To characterize the effects of microstructure, composition and cell seeding density on the kinetics of colonization and remodeling of polysaccharide-based grafts using the perfusion system developed in Specific Aim 2.

The successful attainment of the stated goals will have a significant impact on the treatment of both coronary infarction and deep vein thromboses by reducing or eliminating the need for repeat procedures to clear or replace blocked or stenosed conduits.

Anne M. Mayes, Ph.D.
Associate Professor, Department of Materials Science and Engineering
Massachusetts Institute of Technology (97-0196)
Cambridge, MA

Creation of Ligand-Modified Biomaterials via Entropically-Driven Polymer Segregation

Cell interactions with biomaterial surfaces govern the performance of all implantable devices. Delineation of the molecular players--adhesion receptors and their ligands--has provided a foundation for rational molecular design of surfaces to control cell adhesion, migration, and differentiation by appropriate presentation of ligands on a biomaterial surface. The feasibility of using known small peptide ligands for controlling cell behavior has been demonstrated in model systems such as self-assembled monolayers of alkanethiols on gold. Application of these concepts to materials suitable for human implants remains a formidable challenge.

The objective of the proposed research is to develop a facile method for fabricating biomaterial surfaces that exhibit controlled adhesion, growth, and migration of cells. Interactions between the surface of implanted materials and host tissue are a dominant factor in the effectiveness of biomedical devices, from blood-contacting devices to artificial hips to tissue engineering scaffolds. Control over cell-specific groups present at the surface coupled with protein adsorption resistance makes tailored cell responses possible.

The specific aims are:

1. To demonstrate a new method of surface modification--generalizable to current implants and to those now under development--which renders the surface inert in body fluids.
2. To control the overall density as well as local spatial distribution of ligands that promote cell-specific interactions on a biomaterial surface.
3. To further elucidate the relationship between cell behavior and ligand spatial distribution.

This work will provide a way to treat the surfaces of real biomaterials and devices, i.e., devices made from polymethyl methacrylate, polylactic acid, dacron, etc., so that they inhibit all non-specific protein adsorption and present adhesion ligands. The proposed approach provides systematic control over the density and spatial organization of the adhesion ligands, required to control cell behaviors, using processing approaches readily adaptable to many of today's implants, as well as implants under development.

Moriel NessAiver, Ph.D.
Assistant Professor, Department of Radiology
University of Maryland at Baltimore (97-0201)
Baltimore, MD

Application of Opto-Electronic Implementation of the 2D Discrete Fourier Transform to Real Time Reconstruction of Non-Rectilinear MRI Data Sets

A common feature of all Magnetic Resonance Imaging (MRI) techniques is that they all produce massive amounts of data at very high rates. A single fMRI scan can produce more than 4,000 images in just two minutes, 20-40,000 images in a complete study. Currently, the fastest digital reconstruction systems can just barely keep up with the data acquisition rates and only by using simplified reconstruction algorithms. For the last two years, the PI has been working with an optical processor that is capable of performing the generalized DFT on any rapid MRI data set in as little as 12 ms per image, without the need for interpolation. This is more than five times faster than current fastest digital implementation that using interpolation.

The goal of this research proposal is to compare the image reconstruction and processing results obtained with the current interpolation methods with those obtained using the generalized DFT and to then evaluate the potential applications for true real time reconstruction using an optical processor.

The specific aims are to:

1. Compare image quality obtained using the DFT to that obtained using more common techniques.
2. Perform the same comparisons as in Objective 1 utilizing an optical device.
3. Implement near real-time reconstruction on 1.5T scanner.
4. Demonstrate the value of real-time optical processing.

By providing true real time feedback of potentially higher quality, optical reconstruction could have a positive impact on patient care, both in terms of patient comfort and diagnostic quality of the images, and on patient throughput.

Daniel K. Sodickson, M.D., Ph.D.
Research Associate, Cardiovascular Division
Beth Israel Deaconess Medical Center (97-0252)
Boston, MA

Ultra-Fast Magnetic Resonance Imaging with Radiofrequency Coil Arrays: Techniques, Technologies, and Applications for Partially Parallel MRI

Despite significant advances in imaging speed over the past decade, many diagnostic applications of magnetic resonance imaging (MRI) are still limited by rapid physiologic motion. This research program is aimed at achieving substantial increases in imaging speed using Simultaneous Acquisition of Spatial Harmonics (SMASH), a new strategy for ultra-rapid MRI. Unlike most traditional MR imaging methods which acquire image data in a sequential fashion, the SMASH technique employs a partially parallel data acquisition strategy in which multiple lines of data are generated simultaneously using appropriate combinations of signals from arrays of radiofrequency (RF) coils. Twofold and threefold improvements in imaging speed have been achieved in initial in vivo implementations of SMASH, and significantly larger improvements have been shown in simulations.

Building on the success of initial demonstrations of SMASH imaging, this research program will address several of the central engineering issues associated with scaling up achievable imaging speed without sacrificing image quality. A combination of theoretical investigations, numerical stimulation's, and RF coil array design will be used to remove several of the current technical limitations of the SMASH technique, and to make SMASH a practical tool for ultra-fast clinical MRI. Specific aims of the research program include the following:

1. To characterize signal-to-noise ratio in SMASH imaging.
2. To optimize the handling of RF coil sensitivities in SMASH imaging.
3. To design, construct, and test prototype RF coil arrays for clinical SMASH imaging.
4. To evaluate the performance of the techniques and the apparatus from Specific Aims 1-3 in clinical MR imaging.

The successful outcome of this research will lead to increases in imaging speed by as much as an order of magnitude. Such increases will have application across the spectrum of clinical MRI. Particular advantages are expected in areas such as cardiac MRI, in which rapid cardiac and respiratory motion make imaging speed a primary concern.

J. Thomas Vaughan, Ph.D.
Assistant in Physics, Department of Radiology
The General Hospital Corporation (97-0099)
Charlestown, MA

MR Imaging of the Human Brain at Very High Fields

Increasing the polarizing magnetic field strength for biomedical MR imaging has a number of lures: increased signal-to-noise (SNR), especially for functionally-weighted images, and increased spectral resolution and dispersion for spectroscopic application. Magnet technology has already advanced to make 7T whole body MR systems possible. However, the critical details of radio frequency (RF) energy interactions with the body remain unexplored. Understanding these interactions is crucial both for understanding the safety and practical image quality limits of ultra-high field MR.

The study proposes to quantify field (frequency) dependent RF loss mechanisms, RF field homogeneity, RF power deposition and heating, and the SNR. RF losses will be predicted in analytic and numerical models of the head loaded coil. These coil and tissue losses will in turn be used to predict loss-dependent SNR, magnetic field penetration and RF heating for efficient high field coil designs. These theoretical predictions will be empirically verified by designing, building, and testing prototype high field RF coils with phantom and animal loads. After the feasibility of safe and successful imaging has been determined, 7T human head images will be acquired.

The specific aims are:

1. Quantify RF coil loss mechanisms at field strengths up to 7T.
2. Quantify RF loss mechanisms in tissues at field strengths up to 7T.
3. Quantify RF magnetic field uniformity and its dependence on coil geometry at field strengths up to 7T.
4. Understand and correlate RF power deposition and heating in the head at ultra-high field.
5. Predict and measure SNR dependence on field strength.

A new approach is proposed, from theory to design and implementation, for new MR coils capable of operating efficiently at frequencies up to 300MHz; these frequencies and efficiency levels cannot be achieved with current human head and body coil technologies.

D. Geoffrey Vince, Ph.D.
Project Staff, Department of Biomedical Engineering
The Cleveland Clinic Foundation (97-0332)
Cleveland, OH

Effects of Diabetes on the Structural and Material Properties of Atherosclerotic Plaques

Intravascular ultrasound (IVUS) is a new technique which allows clinicians to produce two-dimensional cross-sectional images of coronary arteries. These images clearly demonstrate the vessel lumen and wall boundaries, thus providing the clinician with a basic outline of the internal structure of the atherosclerotic plaque. It is proposed to develop advanced image processing techniques to analyze IVUS images, produce three-dimensional reconstructions of the arteries, and automatically identify the composition of the atherosclerotic build-up inside the vessel.

There is extensive evidence in the literature that the mechanical properties (such as vessel stiffness) change dramatically in diabetic patients. It is proposed to us IVUS to gather data on the mechanical properties of the vessel by performing vascular compliance measurements. Traditional methods of mechanical testing will also be performed. Data obtained from these tests will be used to produce mathematical models to predict the behavior of arterial plaques. Although mathematical models cannot reproduce the active dynamics of an artery in living patients, they help to integrate and better understand the correlation between pathology observations, IVUS imaging, and mechanical testing. Mathematical modeling can thus be used to predict the role that abnormalities in the arterial wall and changes in plaque composition play in the progression of diabetic atherosclerosis.

The specific aims are:

1. Develop automated 2D and 3D imaging methods.
2. Validate the 2D and 3D imaging techniques.
3. Mechanical testing of blood vessel wall.
4. Mathematical modeling of atherosclerosis.

The applicant envisages that this novel approach to the study of diabetes--utilizing IVUS in combination with advanced image processing techniques, histological analysis, and mechanical testing and modeling--will yield important information on the structural properties of diabetic vessels.

Robert S. Cargill, Ph.D.
Assistant Professor, Department of Mechanical Engineering
Georgia Institute of Technology (97-0111)
Atlanta, GA

Functional Alterations in Isolated Neural-Like Cells Exposed to Graded Mechanical Deformation: Tolerances and Mechanism

Traumatic brain injury (TBI) is a subset of head injuries in which the brain itself is injured, independent of injuries to the cranium. The clinical manifestations of TBI are loss of consciousness in the form of concussion (short-term) or coma (long-term), and in extreme cases, death.

The goal of this project is to determine the correlation between a neural cell's loss of functionality and the applied mechanical deformation that produced this state. The loss of functionality will be determined from the cell's inability to conduct an action potential and from its inability to transmit a signal to the next cell. The experiments will be performed in an in vitro model of TBI that has been developed for this specific purpose. The following specific aims detail the experimental approach:

1. Determine the relationship between the deformation of the surroundings of the neural cell to its local membrane deformation using fluorescent microspheres.
2. Determine the relationship between the applied mechanical deformation and alterations in neuronal functionality (in terms of membrane potential) using an epifluorescence video microscopy system and high strain rate cell deformation device.
3. Develop an injury tolerance criterion describing functional injury as a function of the mechanical deformation parameters.
4. Determine the primary mechanism responsible for the changes in membrane potential induced by the local mechanical deformation.