Albert Macovski - US grants

Perform the research, development, and evaluation necessary to detect and quantify lesions in coronary arteries of humans using a minimally invasive system which is capable of resolving a 0.5 mm lesion in a coronary artery with a lumendiameter of 2 mm or less. Research and development, in vitro and in vivo studies, and evaluation of human studies will be performed with a digital subtraction angiography system in conjunction with intravenous administration of contrast agent.

Chest radiography represents the widest used modality in medical imaging because of its high applicability to a multitude of medical problems. However, as now practiced multiple studies have shown that intervening structures and insufficient contrast result in a disturbingly high percentage of missed lesions. Further, many studies require multiple tomograms with their associated increased dose and image artifacts for adequate characterization of lesions. Recent progress by our group in energy-selective digital radiography (EDR) shows great promise in overcoming these problems. By using the information present in the x-ray energy spectrum, the technique has the capability of producing separate images of bone and soft tissue which dramatically improve the visualization of structures in the thorax. Furthermore, the information is extracted based on physically rigorous techniques developed by our group for computed tomography. Our results indicate that it is sufficiently accurate for in-vivo tissue characterization. Using our unique experimental facilities consisting of a modified General Electric 8800 CT scanner with a high speed voltage switching x-ray generator, we have experimentally shown the capabilities of our techniques. We propose to upgrade our clinical system and to carry out clinical trials to rigorously prove our hypothesis that ESDR provides significant improvement over conventional chest radiography in the detection and characterization of chest lesions, especially those obscurred by overlying bone. Given the success of the initial clinical studies and the extensive analytical and experimental efforts that have been made, this system is in an ideal position to initiate rigorous clinical studies. In parallel with the clinical evaluation, we propose a program of engineering improvements which will make the performance of the bone-subtracted chest image comparable to or greater than that of chest tomography. The initial results of these improved techniques appears to be very promising.

A novel magnetic imaging system using time-varying gradients is proposed for the imaging and characterization of atherosclerotic plaque. A sequence of signals are received in the presence of time-varying gradients in all three dimensions. From these signals, high-resolution three-dimensional image information can be derived to provide detailed anatomic information on the geometric structure of the plaque. In addition, these same signals can be further processed to derive specific spectroscopic data on relevant biochemical parameters of the plaque. Using a novel approach to spectral estimation, images are derived of these relevant parameters. Phantoms, animal models and human subjects will be used to verify the derived images with the actual plaque distribution. In vitro NMR will be used with excised plaque to study the spectroscopic characteristics of various forms of plaque and correlate them with the derived spectroscopic images.

For the past three years, in response to an RFA, we have been studying the use of magnetic resonance imaging to diagnose deep venous thrombosis completely noninvasively and without toxicity or ionizing radiation. Deep venous thrombosis remains a major clinical problem involving a wide class of patients. Many of these patients have gone through surgical procedures or have been immobile for periods of time. During the course of the research we came up with a superb candidate for venous imaging. Our technique is completely insensitive to flow in that it images blood based solely on its material properties. This behavior is essential for the imaging of veins, which may be totally occluded by thrombus. Although other methods, such as gradient echo imaging, are successful with relatively slow flows, they cannot deal with complete or almost complete occlusions. This method provides a reliable and robust approach toward the imaging of the entire venous system, independent of pathology. In addition, a variety of methods of separating veins and arteries have been studied and implemented. In the course of this research we have become involved with a number of significant related studies. These include clot characterization in which the magnetic resonance parameters are used to determine significant clot characteristics such as age. Also, we have studied a complementary venous imaging approach we call "slab scan," which exploits wash-in effects in a sequence of image sections. By making each slab relatively narrow the system becomes responsive to extremely slow flows, although not to a complete occlusion. The desirable properties of this approach include robust vein-artery separation and relative immunity to flow artifacts.

magnetic resonance imaging; angiography;

A file server system for an Information Systems Laboratory will be provided for researchers at Stanford University in the Department of Electrical Engineering. This equipment is provided under the Instrumentation Grants for Research in Computer and Information Science and Engineering program. The research for which the equipment is to be used will be in the areas of optical processing, communication channels, signal quantization and compression, and related areas.

This research seeks to exploit the multiple magnetic resonance properties of materials to produce selective projection images. Unlike existing cross-sectional images, these projections will include the entire volume of interest in a single image, making them ideal for screening. They will selectively visualize the material of interest, such as a specific organ or lesion, while cancelling all others. The objectives of this research include: (1) analytic and experimental studies to determine the optimum procedure, using magnetic resonance imaging, for creating an isolated projection image of important materials of interest, while cancelling all others; (2) determination of the data acquisition and data processing parameters that maximize the signal to noise ratio of the resultant selective image; and (3) verification of the analyses using test phantoms. This project will advance substantially the art of magnetic resonance imaging for medical and other purposes.

Although many exciting correlations have been obtained between various NMR spectroscopic components and tumor diagnosis and treatment, this has yet to become a robust clinical tool. In this application we introduce an array of novel, verified techniques which provide high-resolution images of the important spectroscopic components. These images will provide an excellent clinical tool because of their specificity to tumor activity and their ability to be correlated with conventional anatomic images of the same region. They can therefore be readily interpreted as compared to classical NMR spectra of a volume of tissue. The problems of spectroscopic imaging are legion. Sensitivity dominates the scene, since the desired components are many orders of magnitude below that of the water signal. In addition, inhomogeneity, caused by imperfect magnets and magnetic susceptibility variations in the body, can shift and broaden the components of interest. Thusfar the limited use of spectroscopic images have produced images with very coarse voxels, with questionable value, and very long imaging times. We propose a combination of novel proven techniques which will provide the desired spectroscopic images in a robust fashion. These techniques include: water-referencing to correct both the inter-voxel and intra-voxel frequency errors; estimation theory which, in combination with water-referencing, using a priori information to provide the optimum estimate of weak metabolites buried in noise; time-varying gradients which enable a versatile tradeoff between resolution and imaging time; and correlative filtering which provide high- resolution spectroscopic images by providing the correct amount of high spatial frequencies from the high SNR water image. In addition to these generalized techniques, we plan to apply a variety of novel specialized techniques including a multiple quanta excitation sequence which can separate the important lactate component from the large lipid signals, and a system of imaging components with short T2 relaxation times, such as those found in the phosphorous spectrum.

A system of tumor imaging of the abdomen and thorax is proposed. Although MRI has provided excellent soft tissue contrast in a non-invasive non-toxic fashion, the benefits have accrued primarily in those areas which can be successfully immobilized. Thus the head, pelvis, spine, extremities etc. can be stabilized for a few minute scan. Images of the abdomen and thorax are plagued by severe breathing artifacts. We propose to provide these benefits to all areas of the abdomen and thorax by acquiring the data within a breath-holding interval. This is accomplished by a novel system utilizing a set of interleaved square spiral scans of k-space. This unique approach, which has been successfully implemented, acquires the data in a few seconds using conventional MRI imaging systems. The proposed fast imaging system will be enhanced with a number of important features. Tumors are often difficult to visualize because their characteristics are similar to surrounding tissue. The contrast is significantly improving by using very late echoes, but suffer from excess noise. The images can be profoundly improved using a system of "correlative" filtering where a low-noise early echo is used to enhance the high-contrast late echo image. Other important features of the improved tumor images include flow imaging for tumor vascularity, robust separation of water and fat, and self-focusing excitations which enable the imaging of materials with extremely short relaxation times. Tissues such as lung parenchyma, which is normally not visualized, can be imaged with this approach.

In the course of our research on blood imaging (as distinguished from flow imaging) we have devised a system for measuring and/or imaging oxygen saturation in vessels. Our initial results using both in vitro and in vivo studies appear very encouraging. Oxygen saturation could well provide the most significant functional MR image. We acquire data with this technique in normal imaging times and with typical spatial resolution. We anticipate a wide range of clinical applications. In addition to optimizing and clinically verifying this technique, we plan to study the same effect to estimate the oxygen level in organs and tissue capillary beds.

Due to recent breakthroughs in fast imaging techniques and hardware MRI now has the potential to surpass CT for diagnosing abdominal tumors. MRI has two fundamental advantages: it lacks ionizing radiation, and it shows superior soft tissue contrast. The disadvantages of conventional MRI are sensitivity to motion and inferior resolution. The work proposed here directly addresses these challenges. During the first phase of this grant we have developed several new rapid imaging techniques that clearly have the potential to make tumor imaging in the abdomen a viable clinical tool. These include high-speed interleaved spiral sequences that provide the necessary sensitivity, immunity to motion, and contrast for clinical imaging of tumors in the abdomen. This potential has been vividly confirmed after imaging over 50 patients and many more normal volunteers. These techniques include a novel RARE-spiral sequence that promises improved SNR and resolution. In addition we have studied several volumetric sequences in which 3D kappa- space is scanned efficiently and with insensitivity to motion. We have developed several fluoroscopic sequences to enable critical dynamic studies. For uncooperative patients who cannot hold their breath, we have developed a non-breath-held sequence using navigator-based motion detection system that in real time reacquires all data corrupted by excessive motion. In this renewal we propose to continue developing several important abdominal imaging protocols and also several core techniques critical to high speed imaging with time-varying gradients. Ultimately, this body of work will produce a suite of optimized clinical protocols including breath-held T/1 and T/2 weighted sequences, non-breath-held, motion- compensated T/1 and T/2 weighted sequences, and a fluoroscopic sequence. Although these are designed for current hardware, each can be adapted to exploit the capabilities of the next generation of gradient and RF hardware currently being released. The new protocols and core techniques will be optimized using anecdotal clinical scans.

DESCRIPTION (Adapted from Applicant's Abstract): This re-submission for a competitive renewal aims towards continuing to develop and verify MR oximetry techniques for noninvasive assessment of congenital heart disease. The proposed goal will be approached in three stages. First, they propose to perform an initial clinical study to establish that MR oximetry, with straightforward modifications, can already answer significant clinical questions for cooperative patients who have heart defects with stable hemodynamics, typically atrial septal defects. The second stage will address the problem of performing MR measurements in cooperative patients with adverse flow characteristics such as high flow jets. The third stage extends the application of the previously developed techniques to infants and children where the anatomy is smaller, heart rates are faster, and the ability to cooperate is reduced. At each stage, technically more challenging advances in the measurement process will be required and developed progressively. The results of this investigation will significantly enhance the management of children and young adults with congenital heart disease while extending the general utility of MR oximetry.

magnetic resonance imaging; biomedical equipment development; cryoscience; biomagnetism measurement; radiowave radiation; bioengineering /biomedical engineering;

DESCRIPTION (Adapted from Applicant's Abstract): The applicants proposed to build an ultra-low-cost and high-quality prototype Prepolarized MRI scanner for diagnosing head tumors. The overall goal of this research is to develop a new and innovative imaging modality that shows great potential for cancer screening, diagnosis and image-guided treatment. The specific aims are to design and build both a homogeneous, low-field electromagnet and a lowfrequency receiver system optimized for head imaging. The final aim is to integrate these subsystems with our working Prepolarized MRI system and to obtain an image of a normal volunteers head. This research is intended to demonstrate that Prepolarized MRI system contrast and SNR are equivalent to a conventional 0.5 T MRI system; and that this high-quality Prepolarized MRI system will cost less than $50,000 in capital costs. The successful development and integration of this innovative technology will require the application of several principles of engineering and science. For example, a homogeneous electromagnet design method draws from both electromagnetics and linear programming. A low-noise receiver subsystem design designed by the applicants relies on the principles of optimal RF receiver electronics. Pulse sequence design relies on the principles of MRI physics.

DESCRIPTION (Adapted from Applicant's Description): Cartilage injury leading to osteoarthritis is a leading cause of disability in young people. Osteoarthritis, either based on cartilage injury or degeneration, is a leading cause of disability in the United States. Over the last several decades, much progress has been made in understanding cartilage injury and repair. MRI, with its unique ability to non-invasively image and characterize soft tissue, has shown promise in imaging cartilage. The goal of this proposal is to develop new MRI techniques to assess areas of cartilage damage and repair. Beyond simply demonstrating the morphology of an area of cartilage damage or repair, the investigators will use MRI to characterize the type of tissue present, which has implications about the prognosis and success of the repair. The proposal consists of three components. One is the development of the basic MR imaging technology. This will occur in parallel with the second component, i.e. studies of patients undergoing total knee replacement surgery. This will allow the assessment of the new imaging methods on the wide range of degenerative cartilage conditions typically present in such patients. Finally, a study will be performed to test these new methods in assessing and characterizing cartilage in patients who have had cartilage repair therapy three years previously. MRI holds much promise for assessment of cartilage injury and repair. Emerging therapies for cartilage injury require accurate, non-invasive assessment of repair tissue morphology and content. Development of new techniques to do this has the potential to allow rapid assessment of the effectiveness of these new therapies.

DESCRIPTION (Adapted from Applicant's Abstract): This application is a competing renewal. The original project concerned new fast magnetic resonance imaging (MRI) techniques for improved diagnosis of abdominal tumors. Significant progress was made in developing several fast-scan methods that produce high-quality T2-weighted abdominal images in a breath-hold. The focus of this competing renewal is on new MR methods that take the next step beyond tumor detection---the guidance of tumor biopsies and other interventional procedures using MR. The project title has been modified to reflect this change in focus. The aims of this project are to 1) design and implement a real-time, interactive (RTI) MRI system tailored for interventional procedures, 2) develop and integrate into the RTI system new sequences that provide useful image contrast for interventional procedures; and 3) evaluate this new MR interventional system on phantoms and patients. The research plan includes substantial technical development of a versatile RTI MRI platform, studies of improved MR acquisition methods (e.g., steady-state free precession, fast spin echo, and color flow), and systematic tests to validate the performance of the resultant system. A research team with extensive technical and clinical expertise has been assembled for this project. This expertise, along with the state-of-the-art scanners available, make Stanford University an ideal environment to carry out this project.

The objective of this research project is to demonstrate the technical feasibility of an ultra-low-cost form of magnetic resonance imaging (MRI) called Prepolarized MRI for human extremity imaging. Conventional MRI systems cost between 1 million dollars and 3 million dollars. We propose a novel Prepolarized MRI system that uses two inexpensive electromagnets rather than one expensive superconducting magnet. We aim to demonstrate that a complete dedicated extremity Prepolarized MRI scanner imaging can be constructed for less than 50,000 dollars while maintaining excellent image quality. Over the last five years, we have made excellent progress towards these goals, including our first in vivo human wrist images recorded earlier this month. The total capital cost of our 0.43 T Prepolarized MRI wrist imaging system was less than 35,000 dollars, excluding the MRI data acquisition console. Our principal aim is to expand the size of the imaging system to 31 cm to accommodate knees, feet, and elbows. Once basic imaging is achieived on the 31-cm-bore system, we plan several hardware and software improvements to increase the SNR and contrast flexibility. We also plan a monthly study of normal human knee cartilage conspicuity. Two Stanford Radiologists will compare the normal images with gold-standard images obtained from a 0.5 T GE extremity scanner, and gauge whether the previous month's engineering changes actually improved image conspicuity. We hope at the end of this four year grant to have constructed a 0.5 T Prepolarized MRI system with SNR and contrast comparable to a conventional 0.5 T MRI scanner for less than 50,000 dollars in capital costs. If successful, our research could ultimately lead to significantly greater availability of high-quality Prepolarized MRI scanners, which could be sold for less than the cost of an ultrasound scanner.

DESCRIPTION (provided by applicant): In May of 1999, we were awarded an R21 grant to study and develop a prototype hardware configuration for low-cost magnetic resonance imaging called Prepolarized MRI. During the R21 support, we succeeded in building a Prepolarized MRI scanner for less than $30,000 and we succeeded in acquiring in vivo human hand images. The image quality is roughly comparable to a 0.2 T conventional MRI scanner and is rapidly improving. This R33 application is in response to the Development of Novel Technologies for In Vivo Imaging (PAR-07- 707) Phased Innovation Initiative. Here we propose to extend the technology development initiated in the R21 grant to develop an ultra-low-cost prepolarized MRI head scanner. Prepolarized MRI uses two pulsed magnetic fields. First a strong but inhomogeneous polarizing magnet is pulsed on to polarize hydrogen. This field is turned off and a very low field MRI acquisition follows. A prepolarized MRI scanner with a 0.5 T polarizing field should obtain the image quality of a conventional 0.5 T MRI scanner for a small fraction of the capital cost of a conventional MRI scanner. Here we propose to build a 0.5 T polarizing magnet and a low-field imaging MRI scanner for head imaging. We also plan to develop all the standard imaging pulse sequences that are currently used for detecting and following brain tumors. We plan to investigate the new contrast mechanisms that prepolarized MRI can offer. And finally, we plan to do a small study comparing image quality on a conventional 0.5 T GE MRI scanner. The results of this study will provide feedback for optimizing both the hardware and pulse sequences.