This section deals with areas of great cur-rent interest. Ultrasound provides an excellent framework for observing the interaction of users and developers of technical advances. The major medical uses of ultrasound have appeared within the last 20 years, with exponential growth in the last five to seven years. This paper reviews in a broad sense current applications of ultrasound for cardiac imaging. It will attempt to highlight the need for the amazing devices described in the papers that follow and will mention some areas that await further major advances.

Two-dimensional real time systems have been introduced recently as one of the most significant steps forward in the enlargement of the diagnostic power of ultrasound systems, particularly in cardiology. Although very powerful, especially in the immediate qualitative assessment of cardiac size, geometry, and dynamic behaviour, the present systems as yet have not met the image requirements necessary for easy qualitative analysis. It seems certain that at a future stage research will enable improvement of present day images. The problem areas that remain unsolved are mainly a result of physical limitations.

Instrumentation for echocardiography has progressed through the first generation with single-transducer TM-mode scanning, into the second generation with real-time B-mode imaging using linear arrays of transducers. A third generation with great future potential is represented by two-dimensional piezoelectric transducer arrays which are inherently capable of imaging in a three-dimensional volume as illustrated schematically in Fig. 1. The volume is shown divided into basic depth elements that project through the region with width equal to the lateral resolution. Depth information is obtained by insonifying one or more depth elements with acoustic pulses and receiving echoes from reflecting and scattering objects. Practical methods of displaying real-time three-dimensional information are unavailable, and consequently the most significant capability is the electronic selectability of image data in three dimensions. A scanned system, where electronically selected depth elements are processed sequentially using the same transducers for both insonification and echo reception, is particularly well suited to this use.

Echocardiography is a widely-used noninvasive diagnostic technique that is extremely valuable in evaluating patients with suspected heart disease. As routinely used, this technique utilizes a single ultrasound transducer that can be oriented so that the ultra-sound beam is directed through a specific region of the heart. With this method, rapidly moving structures can be easily visualized. The spatial orientation of cardiac structures, however, is more difficult to evaluate, a limitation that is particularly important when evaluating congenital heart disease where normal anatomic relations cannot be relied upon. In order to overcome this difficulty, several systems have been developed to image a larger region of the heart. These systems can be grouped into a) real-time systems (1-3); i.e., those that allow visualization of heart motion while it actually is occurring and b) non-real time systems (4,5); i.e., those that require multiple consecutive heart beats in order to construct images of the heart.

In 1973 a real-time ultrasonic imaging system, developed at the Stanford Research Institute (SRI), was reported (1). Since that time this laboratory instrument has been improved and modified, and a second instrument, designed expressly for clinical use, has been constructed. The results of preclinical experiments and of the first clinical applications with these ultrasonic cameras have been reported (2-5).

Echocardiograms are commonly used in clinical medicine to diagnose valvular disease by measurement of valve velocity and displacement (1), to determine the presence of pericardial effusions (2), and to measure systolic and diastolic left vertricular diameter (3). The latter measurements may be used to calculate ventricular volume, stroke volume, ejection fraction, and cardiac output under certain defined conditions (4). Standard practice for recording echocardiographic images has been to photograph an oscilloscope with a Polaroid camera. More recently, a cathode ray tube oscillograph has been used to make a continuous record of valve motion, pericardium, and the size of intracardiac chambers as the aim of the transducer is varied. Quantitative measurement of size and motion of cardiac structures is performed using pencil and ruler or, in a few cases, using electronic coordinate digitizers and either computers or desk calculators. The information that can be obtained from a standard echocardiogram is thus limited to that which can be obtained from visual impressions and measurements of photo-graphic records.

Standard one-dimensional, time-motion echocardiography has been extraordinarily useful in the assessment of certain cardiac disorders. This ultrasound technique, unfortunately, is unable to supply detailed information concerning cardiac spatial geometry. As a result, several two-dimensional ultrasound imaging systems have been developed for cardiac use over recent years.

Blood flow is one of the most important physiological parameters, and also one of the most difficult to measure accurately. This is especially so when the flow through a particular vessel or organ is required, either on a chronic basis in experimental animals, or non-invasively in the clinical environment. It now appears that the use of Doppler ultrasound can solve such a measurement problem in a wide variety of cases.

Doppler ultrasound has become the most useful and versatile noninvasive technique for objective evaluation of clinical vascular disease. Commercially available continuous-waxe instruments provide qualitative and quantitative assessment of venous and arterial disease. Pulsed Doppler ultrasound has been developed to provide unique longitudinal and transverse cross-sectional images of the arterial lumen with a resolution approaching that of conventional x-ray techniques. This paper will review our current and anticipated application of Doppler ultrasound in venous, peripheral arterial, and cerebrovascular diseases.

The goals of the various processing and display techniques are to derive from the Doppler spectrum a meaningful measure of the flow phenomenon being investigated and to present this measure to the operator. Important measures go beyond indication of velocity of flow or volume flow to include indications of the type of flow, i.e., laminar or turbulent, as well as geometric parameters such as the depth, width or cross-sectional area of the flow stream.

Much remains unknown about the patterns of ventricular contraction. in healthy and dis eased conditions, particularly for man and animals in an awake unanesthetized state. Over the past 80 years roentgenologic methods, particularly fluoroscopic observations, remain. the simplest. and most practical methods for obtaining functional information. about the heart. With advances in radiographic technology, various elaborate methods have been developed to record precise and detailed informa tion about. heart functions utilizing conven tional large-film radiography and roentgen video- and cinematography. (1). Such studies have allowed. for measurement of cardiac chamber volumes and dimensions and for the description, of cardiac function, in hydraulic and mechanical terms (2,3). The portionofthis book dealing with anglocardlography and image processing is divided into three parts: (1) a. discussion. of the available instrumentation, (2) an evaluation of methods for extracting and processing data from radiological images, and (3) an assessment of clinical uses of derived. information. It will be the purpose of this introductory presentation briefly to review the history of angiography and the manner in which it has been used to make qualitative and quantitative measurements of heart and vessel geometry. Resultant. information coll cerning internal cardiac structure and form has added, a new depth to assessment of physio logic function and an improved ability to evaluate cardiac performance duringdiseased states. Automation of measurements is now proceeding. rapidly. As these, methods are proving to be of value they are being applied increasingly to larger sections of the population to assist in cardiac diagnosis and health care.

Biplanar high-speed cineradiography a technique that offers a unique approach for obtaining precise physiologic data that extends beyond the conventional recording devices. The rapid frame rates reduce motion blurring of moving objects and when used in the biplanar mode allows the inves-tigator to examine this information in three dimensional space. In our laboratory we have found it to be a useful technique for examining cardiovascular dynamic changes, particularly those events which are difficult or impossible to describe or measure using standard cine or physiologic recording methods. " " When combined witha number of new radiopaque contrast materials the investigator can obtain in-formation about the changing size and shape of the cardiac chambers throughout the heart cycle; the patterns of flaw through these chambers and the surrounding great vessels; the motion of the heart valves and the correlation of this data with pressures, sounds or electrical events. This type of information enables the scientist working with a small laboratory computer to calculate rapidly changing forces, volumes, dimenfpns and flaws in the cardiovascular system.

The first and second generation computer assisted cross-sectional reconstruction systems, such as the EMI, ACTA and DELTA brain and whole-body scanners, because of their excellent density resolution have produced accurate cross-sectional reconstructions of the anatomical structure of stationary organs, particularly the brain, the latter of which have revolutionized clinical neuroradiology (1-4).

Considerable effort has been devoted in recent years towards developing indices of myocardial performance based on the contractile geometry of the left ventricle. Beginning with the pioneer work of A.V. Hill (1), the force-velocity-length characteristics of isolated skeletal and cardiac muscle have been studied extensively (2-4). Application of this knowledge to the evaluation of the intact heart presents many problems. Angiographic measurements are subject to a variety of technical errors. In addition, there exists no exact model relating performance of the intact heart to the behavior of individual muscle fibers. The geometry of the ventricle is complex and difficult to account for accurately. Despite these difficulties, angiographic estimates of contractility are useful in evaluating myocardial performance.

Regional patterns of sequential contraction of the right and left ventricles under physiological conditions, have been described (1,2) and the important detrimental role of abnormal asynchronous contraction patterns in the heart due to aberrant electrical activation or ischemic regions has also been recognized. Such data have indicated the necessity for measurements of regional myocardial function over a major portion of the ventricular wall but for practical technological reasons most geometric measurements made directly on the dynamic intact heart involve a small number of regions of the ventricle. Simplifying shape assumptions such as those commonly used in calculations of ventricular volumes (3,4) or myo-cardial wall mass (5), as well as the arbitrary selection of frames of reference in the quantitation of regional dynamics all affect the validity of these indices to an unknown extent.

Dynamic measurements of the size and shape of the left ventricular cavity and their corre-lation with simultaneously occurring pressure and flow events are of great importance in cardiovascular research. Many methods are known for measuring cardiac chamber dimensions and volume in animals and man. However, angio-cardiography has proven with time to be the most readily available and reliable method for these purposes (1).

The capabilities of x-ray fluoroscopic imaging systems to provide large amounts of quantitative information in both clinical procedures and in a research environment has not been generally appreciated. In addition, in many instances the chain can be modified to produce single images at substantially lower patient dose than when operated in the conventional manner. Even when operated in the conventional manner, the analog video signal is capable of providing much physiological information by utilizing the techniques of video densitometry (1-5). In addition the analog signal has been processed by computer to yield quantitative data for certain cardiac functions (6-9). However, modification to single scan operation provides images with lower patient dose and with higher spatial resolution. Conversion to a digital format provides the opportunity for use of simple processing for the elimination of statistically independent noise thereby increasing precision, absolute accuracy, and gray scale sensitivity.

A computer image processing technique has been developed to estimate the degree of athero sclerosis in the human femoral artery. With an angiographic film of the vessel as input, the computer has been programmed to estimate vessel abnormality through a series of measurements, some derived primarily from the vessel edge in formation and others from optical density vari ations within the lumen shadow. These measurements are combined into an atherosclerosis index, which as described below was found in a post-mortem study to correlate well with both visual and chemical estimates of atheroscle rotic disease.

The problem of digitally reconstructing the image of the internal structure of an object from measurements of its two dimension-al projections, resulting from transmission of radiation through the object, has been of interest in mathematics, radio-astronomy, and biological sciences for over 50 years. About four years ago this problem suddenly assumed major importance in medical engineering after Houncefield and his associates announced the development of a sophisticated computerized x-ray scanning system capable of reconstructing cross-sectional images (tomograms) of the human head with high tissue-density discrimi-nation capability [23]. This development was heralded as a major breakthrough in diagnostic radiology because it, for the first time, permitted non-invasive visualization of the ventricles of the brain, as well as a large class of brain tumors and injuries. More recent research has demonstrated the success of this type of imaging with other parts of the human body.

The Cardiovascular Research Laboratory at Ames Research Center is concerned with heart performance under conditions of stress associated with flight. Physiological studies are conducted on human and animal subjects using hemodynamic and angiographic techniques to assess cardiac function. Studies over the past few years have demonstrated the importance of size and shape changes in the cardiac ventricle in reflecting changes in physiological performance at rest or following interventions such as drugs, exercise, or surgery (1, 2). Experience has also shown that angio-cardiography is the only reliable means of measuring overall changes for the chamber under study. Although many other means are available for obtaining heart dimensional information, they usually provide only a single dimension such as a chamber diameter. This paper surveys our methods for accurate measurement and meaningful display of cardiac dimensions as obtained from fluoroscopy. The emphasis has been on scaling recorded images back to actual dimensions and displaying cardiac motion to assess functional changes. This has been accomplished by use of a low-cost but flexible ensemble of computer and video equipment which assists in (1) digitizing image outlines or points, (2) filing digitized images for later recall in any sequence, (3) correcting recorded images for distortion, (4) modeling heart geometry, (5) analyzing dimensional changes, and (6) displaying both heart geometry and results of analysis in a useful form.

The measurement of regional myo-cardial dynamics and perfusion over the entire anatomic and temporal extents of the ventricular walls and successive cardiac cycles, respectively, is becoming widely recognized (1-5) as a necessary requirement for obtaining accurate assessments of cardiac function. Similarly, determinations of the true dynamic regional changes in shape and dimensions of the lungs are necessary for accurate analyses of pulmonary function.

The growing number and success of computer-aided measurement techniques both within the medical world (cell counting) and outside of it (processing bubble chamber scans) suggest the arrival of a sensory extension dependent upon computers. Whether this exten-sion can be made cost-effective or must remain a laboratory toy is being investigated in many independent efforts.

The increasing use of left ventricular cineangiography as a diagnostic and decision-making tool in patients with coronary heart disease has resulted in the definition of certain gross regional abnormalities of wall motion, principally akinesis and paradoxical motion or dyskinesis. These abnormalities have been shown to correlate well with sites of coronary obstructive lesions and with electrocardiographic evidence of myocardial infarction (1,2,3). Such areas of abnormal wall motion may be best displayed by the superimposition of traced left ventricular silhouettes but they are usually obvious on viewing the projected cineangiograms.

Quantitative angiographic analysis has been recognized as an important method for defining the hydraulic function of the heart. A video disc light-pen computer system has been devel-oped at Stanford Medical Center in order to facilitate quantitation of ventriculographic images (1). Development of this system was prompted by the need to provide measures of overall left ventricular pump performance on a routine basis for clinical decision making and to provide a method for implementation of new experimental approaches. This system provides the capability for computation of ventricular volumes and other measurements from operator-defined margins by greatly reducing the tedium and errors associated with manual planemetry.

The systolic ejection fraction has been the standard method for assessing left ventricular performance. However, it is evident that this measure of overall ventricular performance does not fully describe abnormalities in local myocardial performance characteristic of coronary artery disease. For example, areas of hypokinesis can frequently be found in subjects with normal ejection fraction. Several methods have been proposed for the study of local wall motion based on the change in the length of chords drawn to the endocardial silhouette from left ventricular single or bi-plane angiograms (1,2,3,4,5,6,7). More recently, Welch has suggested that the extent of local wall motion at midsystole is more often abnormal than the total shortening at end sys-stole in ischemic heart disease (8).

The standard clinical practice of estimating blood flow in individual coronary arteries from subjective angiographic evalua-tion of the severity of stenosis in these vessels is fraught with error. Most current clinically available methods of determining regional myocardial flow are technically complex and expensive, or limited in resolution, and in simplicity and convenience to the clinician/investigator and his patient.

A clinical program designed to evaluate the effects of risk factor reduction on the progression of atherosclerosis was initiated at the University of Southern California in 1971. Since atherosclerosis is, at least partially, reversible in the experimental animal(1,2,3), it is possible, although not proven, that regression may occur in man. Classical epidemiologic studies characterize only progression of the disease, since end-points are appearance of end-organ damage or death. Development of a method for the measurement of atherosclerosis in the living individual is essential in a human regression study. The method would have to: 1. Be reproducible 2. Detect early non-obstructive lesions. 3. Be amenable to automated grading methodology.

Almost 50 years have passed since Blumgart and Weiss first used a cloud chamber to follow a radium salt solution through a patient's heart (1). Extensive advances in the technology of the detectors, data analysis systems, and tracers used have resulted in greatly expanded applications of radioisotopes to the assessment of cardiac function and disease. The development of nuclear cardiology has proceeded along four lines: 1) radionuclide angiography, 2) myocardial perfusion imaging, 3) intracoronary microsphere imaging, and 4) regional myocardial blood flow determination using inert gases.

More than 100,000 Americans die each year from fatal heart attacks. Whereas, coronary intensive care units have made significant progress in controlling potentially lethal arrhythmias, there has been no success in reversing cardiogenic shock. Methods proposed for treating patients with myocardial infarction or severe myocardial ischemia include early diagnosis and chemical or physical in-tervention measures designed to minimize the volume of the infarcted tissue. In order to evaluate these approaches, we seek new meth-ods for noninvasively quantitating the volume of the ischemic and infarcted myocardium. Some isotopes which localize in the healthy myocardium, and others which accumulate specifically in the ischemic myocardium, offer a sensitive noninvasive technique of following myocardial injury. However, simple two-dimensional imaging of the distribution of injected isotopes such as 43K, 81Rb, 201T1 for negative imaging, or 99mTc-tetracycline, 99mTc-pyrophosphate, or 203Hg-chlormerodrin for positive imaging, is not adequate for quantitative and definitive evaluation of the function of the myocardium. Gamma camera or rectilinear scanner images suffer from the fact that the isotope distribution throughout the entire thorax is projected onto a plane and the resulting poor contrast image repre-sents not only myocardial uptake, but appreciable uptake by skeletal muscle and adjacent soft tissues for negative agents such as 201T1; and uptake by the ribs and sternum for the positive agents such as 99mTc-pyrophos-phate. These problems can be circumvented by three-dimensional imaging techniques, and this paper is devoted to a discussion of three methods of imaging the three-dimensional dis-tribution of isotopes in the myocardium: (1) Three-dimensional imaging using multiple Angercamera views. (2) Longitudinal tomographic images with compensation for blurring. (3) Transverse-section reconstruction using coincidence detection of annihilation gammas from positron emitting isotopes.

Several indices of left ventricular systolic performance have been proposed, one of which, left ventricular ejection fraction, has been extensively applied to clinical problems of patients with heart disease. Left ventricular ejection fraction, which is the ratio of left ventricular stroke volume to end-diastolic volume, is usually performed by left heart catheterization with contrast ventriculography. Catheterization ordinarily requires hospitalization, is relatively expensive and has a definite, although low, morbidity and mortality. In particular contrast ventriculography does not lend itself to serial study at frequent intervals. Consequently several radionuclide methods have been developed which accurately measure left ventricular ejection fraction. In these methods an Anger scintillation camera is used with 99mTechnetium to record either an electrocardiographically gated (1) or a dynamic radionuclide angiocardiogram (2)(3).

The conventional methods for the assess-ment of myocardial functional integrity, which rely on the external detection of an administered radiopharmaceutical, share the following crippling limitations. (1) The field of view of the radiation detector always encompasses the contribution of activity contained in tissues overlying and underlying the plane of interest. This contribution, which in general cannot be subtracted from the measured signal, contributes noise, and in general considerably depresses the contrast in the plane of interest between tissues containing different activities. (2) The attenuation of the gamma radiation originating from the activity in the myocardium is dependent upon the amount of tissue interposed between the myocardium and the radiation detector. This effect is difficult to compensate for. The combination of the above two defects seriously depresses contrast in myocardial nuclear imaging, and all but precludes the reliable quantitative assessment of the distribution of a radio-active label in the myocardium.

The recognition of acute myocardial infarction in the past has depended on certain electrocardiogram and cardiac enzyme alterations in association with a particular clinical history. However, in certain instances these parameters are not sufficient to either make a positive diagnosis of acute myocardial infarction or to exclude its presence. In particular, the presence of left bundle branch block, previous myocardial infarction and recent cardioversion all provide certain difficulties as far as the electrocardiographic recognition of acute myocardial infarction is concerned. Moreover, subendocardial myocardial infarction can never be precisely identified from the electrocardiogram alone; this is a matter of some concern since the subendo-cardium is the area most vulnerable to ischemic damage in experimental animals (1-3). In general, cardiac enzyme abnormalities are not so specific for myocardial damage that enzyme elevations in any individual patient might not be due to congestive heart failure, hemolysis, brain or pulmonary damage, intramuscular injection, etc., although the recent studies utilizing the myocardial specific creatine phosphokinase isoenzyme have suggested that it may be of more value in the exact recognition of myocardial infarction than the others (4).

There is clearly a need for non-invasive techniques that will permit the study of the changes in left ventricular performance in the acutely ill patient early after an acute myocardial infarction as well as during the recovery period. Left ventricular ejection fraction as a measure of left ventricular performance can be obtained noninvasively by echocardiography.1/41) It is derived by measuring a single chord in the transverse plane of the left ventricular chamber. For the calculation of ejection fraction by this method two assumptions are made: (1) the geometry of the left ventricular chamber can be approximated by an elliptical model; and (2) the recorded transverse segmental function is representative of the entire left ventricle. The accuracy of this technique, therefore, might be limited when regional abnormalities in left ventricular wall motion are present, e.g., in patients with myocardial infarction. Alternatively, radioisotopic high-frequency left ventricular timeactivity dilution curves provide a means to measure left ventricular ejection fraction with high accuracy. (2,3,4) The calculation of the ejection fraction by this method is virtually free of geometric assumptions and, therefore, should be par-ticularly useful in patients with myocardial infarctions.

Throughout this text many new methods and technologies for obtaining and processing images of the cardiovascular system have been presented. Many of these techniques and instruments are the most advanced available and are at the frontier between what is now commercially available and what will be offered to the general medical community during the next five years. Obviously, for any new device to make a broad impact on treating the everincreasing number of patients with cardiovascular disease, it must be manufactured and distributed widely. The mechanisms by which wide distribution occurs raise many questions. How does a new medical device enter the commercial market? What are the important questions raised by prospective manufacturers? What factors help to determine whether a new instrument will make a profit for a company? And finally, by what criteria can the consumer evaluate the large number of medical devices on the market and make a logical decision regarding which one will meet his long range needs?

The National Aeronautics and Space Administration(NASA) has assumed responsibility not only for America's Space Program, but also for insuring that the technological advances resulting from space exploration are fully utilized for the public's benefit. In order to provide an effective mechanism for transferring aerospace technology to the biomedical field, four Biomedical Application Teams (BATeams) have been established by NASA. Their purpose is to identify, validate, and disseminate aerospace technology which could have significant impact on biomedical research and clinical medical practice. Each BATeam has programs focusing on specific medical problem areas. The team at Stanford operates both through the division of Cardiology and the NASA-Ames Research Center in nearby Mountain View. This team of physicians and engineers emphasizes NASA-developed techniques and instruments which have direct application to the diagnosis and treatment of cardiac diseases.

Industry does a lot of good things in our society. It provides jobs to many workers. It produces goods and services, including many valuable new products. It pays taxes that help support such useful enterprises as NASA, the National Institutes of Health and Stanford Medical School. (Corporate income taxes came to about $56 billion in 1974.) And it pays dividends to stockholders which include pension funds for professors, policemen, and other meritorious persons.

In order to synthesize the view of any technology transfer and associated enterprise from a venture capital standpoint it is necessary to first know what the objectives of venture capital are.

In the case of the ultrasonoscope, agree with Professor Enthoven's analysis and his conclusTon. The product is not significantly better than others already on the market to offer a competitive advantage worthy of the investment and uncertainties associated with the process of bringing it to market. Clinical benefits such as portability, safety and flexibility are important in their own right, but seem marginal here in competitive product differentiation. And, for a company without an established position in the field, these are not the sort of contributions that make for successful market entry.

During the past two decades, new instrumentation has been introduced into cardiology with increasing frequency. In many instances this is the result of the clinical application of technology developed in other scientific fields and then applied to specific medical, diagnostic and therapeutic problems. Although a modicum of success has been achieved in introducing these new technologies in cardiology, the potential for even greater technological advances remains large. However, problems occurring between the producer or developer of new cardiologic technology and the user have become more complex. The potential user is faced frequently with deciding whether or not to accept a new piece of equipment or a new technological concept into the practice of cardiology. In my discussion I will focus my attention on new equipment which permits imaging of the cardiovascular system and will outline my own approach for evaluating these new instruments before introducing them into the clinical arena.