A RevolutionGraphical Programming and Virtual Instrumentation
88.2 Virtual Instrumentation and Biomedical Engineering
Example 1 Example 2
88.1 Overview
A RevolutionGraphical Programming and Virtual Instrumentation
Over the last decade, the graphical programming revolution has empowered engineers to develop cus- tomized systems, the same way the spreadsheet has empowered business managers to analyze nancial data. This software technology has resulted in another type of revolutionthe virtual instrumentation revolution, which is rapidly changing the instrumentation industry by driving down costs without sacricing quality. Virtual Instrumentation can be dened as: A layer of software and/or hardware added to a general-purpose computer in such a fashion that users can interact with the computer as though it were their own custom-designed traditional electronic instrument. Today, computers can serve as the engine for instrumentation. Virtual instruments utilize the open architecture of industry-standard computers to provide the processing, memory, and display capabilities; while the off-the-shelf, inexpensive interface boards plugged into an open bus, standardized communi- cations bus provides the vehicle for the instruments capabilities. As a result, the open architecture of PCs and workstations allow the functionality of virtual instruments to be user dened. In addition, the processing power of virtual instruments is much greater than stand-alone instruments. This advantage will continue to accelerate due to the rapid technology evolution of PCs and workstations that results from the huge investments made in this industry. The major benets of virtual instrumentation include increased performance and reduced costs. In addition, because the user controls the technology through software, the exibility of virtual instrumentation
Eric Rosow
Hartford Hospital and Premise Development Corporation
Joseph Adam
Premise Development Corporation
2000 by CRC Press LLC
is unmatched by traditional instrumentation. The modular, hierarchical programming environment of virtual instrumentation is inherently reusable and recongurable.
88.2 Virtual Instrumentation and Biomedical Engineering
Virtual Instrumentation applications have encompassed nearly every industry including the telecommu- nications, automotive, semiconductor, and biomedical industries. In the elds of healthcare and biomed- ical engineering, virtual instrumentation has empowered developers and end-users to conceive of, develop, and implement a wide variety of research-based biomedical applications and executive infor- mation tools. These applications fall into several categories including: clinical research, equipment testing and quality assurance, data management, and performance improvement. In a collaborative approach, physicians, researchers, and biomedical and software engineers at Hartford Hospital (Hartford, CT) and Premise Development Corporation (Avon, CT) have developed various data acquisition and analysis systems that successfully integrate virtual instrumentation principles in a wide variety of environments. These include: The EndoTester, a patented quality assurance system for beroptic endoscopes a Non-Invasive Pulmonary Diffusion and Cardiac Output Measurement System a Cardiovascular Pressure-Dimension Analysis System BioBench, a powerful turnkey application for physiological data acquisition and analysis PIVIT, a Performance Indicator Virtual Instrument Toolkit to manage and forecast nancial data a Virtual Intelligence Program to manage the discrete components within the continuum of care BabySave, an analysis and feedback system for apnea interruption via vibratory stimulation This chapter will describe several of these applications and describe how they have allowed clinicians and researchers to gain new insights, discover relationships that may not have been obvious, and test and model hypotheses based on acquired data sets. In some cases, these applications have been developed into commercial products to address test and measurement needs at other healthcare facilities throughout the world.
Example 1: BioBenchA Virtual Instrument Application for Data Acquisition and Analysis of Physiological Signals
The biomedical industry is an industry that relies heavily on the ability to acquire, analyze, and display large quantities of data. Whether researching disease mechanisms and treatments by monitoring and storing physiological signals, researching the effects of various drugs interactions, or teaching students in labs where students study physiological signs and symptoms, it was clear that there existed a strong demand for a exible, easy-to-use, and cost-effective tool. In a collaborative approach, biomedical engi- neers, software engineers and clinicians, and researchers created a suite of virtual instruments called BioBench. BioBench (National Instruments, Austin, TX) is a new software application designed for physiolog- ical data acquisition and analysis. It was built with LabVIEW, the worlds leading software development environment for data acquisition, analysis, and presentation.
1
Coupled with National Instruments data acquisition (DAQ) boards, BioBench integrates the PC with data acquisition for the life sciences market. Many biologists and physiologists have made major investments over time in data acquisition hardware built before the advent of modern PCs. While these scientists cannot afford to throw out their investment in this equipment, they recognize that computers and the concept of virtual instrumentation yield tremendous benets in terms of data analysis, storage, and presentation. In many cases, traditional
1
BioBench was developed for National Instruments (Austin, TX) by Premise Development Corporation (Avon, CT).
2000 by CRC Press LLC
medical instrumentation may be too expensive to acquire and/or maintain. As a result, researchers and scientists are opting to create their own PC-based data monitoring systems in the form of virtual instruments. Other life scientists, who are just beginning to assemble laboratory equipment, face the daunting task of selecting hardware and software needed for their application. Many manufacturers for the life sciences eld focus their efforts on the acquisition of raw signals and converting these signals into measurable linear voltages. They do not concentrate on digitizing signals or the analysis and display of data on the PC. BioBench is a low-cost turnkey package that requires no programming. BioBench is compatible with any isolation amplier or monitoring instrument that provides an analog output signal. The user can acquire and analyze data immediately because BioBench automatically recognizes and controls the National Instruments DAQ hardware, minimizing conguration headaches. Some of the advantages of PC-Based Data Monitoring include: Easy-to-use-software applications Large memory and the PCI bus Powerful processing capabilities Simplied customization and development More data storage and faster data transfer More efcient data analysis Figure 88.1 illustrates a typical setup of a data acquisition experiment using BioBench. BioBench also features pull-down menus through which the user can congure devices. Therefore, those who have made large capital investments can easily migrate their existing equipment into the computer age. Integrating a combination of old and new physiological instruments from a variety of manufacturers is an important and straightforward procedure. In fact, within the clinical and research setting, it is a common requirement to be able to acquire multiple physiological signals from a variety of medical devices and instruments which do not necessarily communicate with each other. Often times, this situation is compounded by the fact that end-users would like to be able to view and analyze an entire waveform and not just an average value. In order to accomplish this, the end-user must acquire multiple channels of data at a relatively high sampling rate and have the ability to manage many large data les. BioBench can collect up to 16 channels simultaneously at a sampling rate of 1000 Hz per channel. Files are stored in an efcient binary format which signicantly reduces the amount of hard disk and memory require- ments of the PC. During data acquisition, a number of features are available to the end-user. These features include:
FIGURE 88.1
A typical biomedical application using BioBench (courtesy of National Instruments).
2000 by CRC Press LLC
Data Logging:
Logging can be enabled prior to or during an acquisition. The application will either prompt the user for a descriptive lename or it can be congured to automatically assign a lename for each acquisition. Turning the data logging option on and off creates a log data event record that can be inspected in any of the analysis views of BioBench.
Event Logging:
The capacity to associate and recognize user commands associated with a data le may be of signicant value. BioBench has been designed to provide this capability by automatically logging user-dened events, stimulus events, and le logging events. With user-dened events, the user can easily enter and associate date and time-stamped notes with user actions or specic subsets of data. Stimulus events are also data and time-stamped and provide the user information about whether a stimulus has been turned on or off. File logging events note when data has been logged to disk. All of these types of events are stored with the raw data when logging data to le and they can be searched for when analyzing data.
Alarming:
To alert the user about specic data values and thresholds, BioBench incorporates user- dened alarms for each signal which is displayed. Alarms appear on the user interface during data acquisition and notify the user that an alarm condition has occurred. Figure 88.2 is an example of the Data Acquisition mode of BioBench. Once data has been acquired, BioBench can employ a wide array of easy-to-use analysis features. The user has the choice of importing recently acquired data or opening a data le that had been previously acquired for comparison or teaching purposes. Once a data set has been selected and opened, BioBench allows the user to simply select and highlight a region of interest and choose the analysis options to perform a specic routine. BioBench implements a wide array of scalar and array analyses. For example, scalar analysis tools will determine the minimum, maximum, mean, integral, and slope of a selected data set, while the array analysis tools can employ Fast Fourier Transforms (FFTs), peak detection, histograms, and X vs. Y plots. The ability to compare multiple data les is very important in analysis and BioBench allows the user to open an unlimited number of data les for simultaneous comparison and analysis. All data les can be scanned using BioBenchs search tools in which the user can search for particular events that are
FIGURE 88.2
BioBench acquisition mode with alarms enabled.
2000 by CRC Press LLC
associated with areas of interest. In addition, BioBench allows the user to employ lters and transforma- tions to their data sets and all logged data can be easily exported to a spreadsheet or database for further analysis. Finally, any signal acquired with BioBench can be played back, thus taking lab experience into the classroom. Figure 88.3 illustrates the analysis features of BioBench.
Example 2: A Cardiovascular Pressure-Dimension Analysis System
Introduction
The intrinsic contractility of the heart muscle (myocardium) is the single most important determinant of prognosis in virtually all diseases affecting the heart (e.g., coronary artery disease, valvular heart disease, and cardiomyopathy). Furthermore, it is clinically important to be able to evaluate and track myocardial function in other situations, including chemotherapy (where cardiac dysfunction may be a side effect of treatment) and liver disease (where cardiac dysfunction may complicate the disease). The most commonly used measure of cardiac performance is the ejection fraction. Although it does provide some measure of intrinsic myocardial performance, it is also heavily inuenced by other factors such as heart rate and loading conditions (i.e., the amount of blood returning to the heart and the pressure against which the heart ejects blood). Better indices of myocardial function based on the relationship between pressure and volume through- out the cardiac cycle (pressurevolume loops) exist. However, these methods have been limited because they require the ability to track ventricular volume continuously during rapidly changing loading con- ditions. While there are many techniques to measure volume under steady state situations, or at end- diastole and end-systole (the basis of ejection fraction determinations), few have the potential to record volume during changing loading conditions. Echocardiography can provide online images of the heart with high temporal resolution (typically 30 frames per second). Since echocardiography is radiation-free and has no identiable toxicity, it is ideally suited to pressurevolume analyses. Until recently however, its use for this purpose has been limited by the need for manual tracing of
the endocardial borders, an extremely tedious and time-consuming endeavor.
FIGURE 88.3
BioBench analysis mode.
2000 by CRC Press LLC
The System
Biomedical and software engineers at Premise Development Corporation (Avon, CT), in collaboration with physicians and researchers at Hartford Hospital, have developed a sophisticated research application called the Cardiovascular Pressure-Dimension Analysis (CPDA) System. The CPDA system acquires echocardiographic volume and area information from the acoustic quantication (AQ) port, in conjunc- tion with vetricular pressure(s) and ECG signals to rapidly perform pressurevolume and pressure-area analyses. This fully automated system allows cardiologists and researchers to perform online pressure- dimension and stroke work analyses during routine cardiac catheterizations and open-heart surgery. The system has been designed to work with standard computer hardware. Analog signals for ECG, pressure, and area/volume (AQ) are connected to a standard BNC terminal board. Automated calibration routines ensure that each signal is properly scaled and allows the user to immediately collect and analyze pressure- dimension relationships. The CPDA can acquire up to 16 channels of data simultaneously. Typically, only three physiological parameters, ECG, pressure, and the AQ signals are collected using standard data acquisition hardware. In addition, the software is capable of running on multiple operating systems including Macintosh, Windows 95/98/NT, and Solaris. The CPDA also takes advantage of the latest hardware developments and form-factors and can be used with either a desktop or a laptop computer. The development of an automated, online method of tracing endocardial borders (Hewlett Packards AQ Technology) (Hewlett-Packard Medical Products Group, Andover, MA) has provided a method for rapid online area and volume determinations. Figure 88.4 illustrates this AQ signal from a Hewlett Packard Sonos Ultrasound Machine. This signal is available as an analog voltage (1 to +1 volts) through the Sonos Dataport option (BNC connector).
Data Acquisition and Analysis
Upon launching this application, the user is presented with a dialog box that reviews the license agreement and limited warranty. Next, the Main Menu is displayed, allowing the user to select from one of six options as shown in Fig. 88.5.
Clinical Signicance
Several important relationships can be derived from this system. Specically, a parameter called the
End- Systolic Pressure-Volume Relationship
(
ESPVR
)
describes the line of best t through the peak-ratio (max- imum pressure with respect to minimum volume) coordinates from a series of pressurevolume loops
FIGURE 88.4
The Acoustic Quantication (AQ) signal (Hewlett Packard).
2000 by CRC Press LLC
generated under varying loading conditions. The slope of this line has been shown to be a sensitive index of myocardial contractility that is independent of loading conditions. In addition, several other analyses, including
time varying elastance
(
E
max
)
and
stroke work,
are calculated.
Time-varying elastance is measured by determining the maximum slope of a regression line through a series of isochronic pressurevolume coordinates. Stroke work is calculated by quantifying the area of each pressurevolume loop. Statistical parameters are also calculated and displayed for each set of data. Figure 88.7 illustrates the pressure- dimension loops and each of the calculated parameters along with the various analysis options. Finally, the user has the ability to export data sets into spreadsheet and database les and export graphs and indicators into third-party presentation software packages such as Microsoft PowerPoint.
FIGURE 88.5
Cardiovascular pressure-dimension analysis main menu.
FIGURE 88.6
The data selection front panel.
2000 by CRC Press LLC
Summary
Virtual Instrumentation allows the development and implementation of innovative and cost-effective biomedical applications and information management solutions. As the healthcare industry continues to respond to the growing trends of managed care and capitation, it is imperative for clinically useful, cost-effective technologies to be developed and utilized. As application needs will surely continue to change, virtual instrumentation systems will continue to offer users exible and powerful solutions without requiring new equipment or traditional instruments.
References
1. Fisher, J.P., Mikan, J.S., Rosow, E., Nagle, J., Fram, D.B., Maffucci, L.M., McKay, R.G., and Gillam, L.D., Pressure-Dimension Analysis of Regional Left Ventricular Performance Using Echocardio- graphic Automatic Boundary Detection: Validation in an Animal Model of Inotropic Modulation,
J. Am. College of Cardiol.,
19(3), 262A, 1992. 2. Fisher, J.P., McKay, R.G., Mikan, J.S., Rosow, E., Nagle, J., Mitchel, J.F., Kiernan, F.J., Hirst, J.A., Primiano, C.A., Fram, D.B., Gillam, L.D., Hartford Hospital and University of Connecticut, Hart- ford, CT,
Human Left Ventricular Pressure-Area and Pressure-Volume Analysis Using Echocardio- graphic Automatic Boundary Detection.
65th Scientic Session of the American Heart Association (11/92). 3. Fisher, J.P., Mitchel, J.F., Rosow, E., Mikan, J.S., Nagle, J., Kiernan, F.J., Hirst, J.A., Primiano, Gillam, L.D., Hartford Hospital and University of Connecticut, Hartford, CT,
Evaluation of Left Ventricular Diastolic Pressure-Area Relations with Echocardiographic Automatic Boundary Detection,
65th Sci- entic Session of the American Heart Association (11/92). 4. Fisher, J.P., McKay, R.G., Mikan, J.S., Rosow, E., Nagle, J., Mitchel, J.F., Fram, D.B., Gillam, L.D., Hartford Hospital,
A Comparison of Echocardiographic Methods of Evaluating Regional LV Systolic Function: Fractional Area Change vs. the End-Systolic Pressure-Area Relation.
65th Scientic Session of the American Heart Association (11/92).
FIGURE 88.7
The cardiac cycle analysis front panel.
2000 by CRC Press LLC
5. Fisher, J.P., McKay, R.G., Rosow, E. Mikan, J. Nagle, J. Hirst, J.A., Fram, D.B., Gillam, L.D.,
On- Line Derivation of Human Left Ventricular Pressure-Volume Loops and Load Independent Indices of Contractility Using Echocardiography with Automatic Boundary Detection-A Clinical Reality.
Comparison of Mechanical and Pharmacologic Methods of Altering Loading Conditions to Determine End-Systolic Indices of Left Ventricle Function.
Circulation 90 (II), 1-494, 1994.
7.
Fisher, JP, Martin, J., Day FP, Rosow, E., Adam J, Chen C, Gillam LD. Validation of a Less Invasive Method for Determining Preload Recruitable Stroke Work Derived with Echocardiographic Auto- matic Boundary Detection. Circulation 92, 1-278, 1995. 8. Fontes ML. Adam J, Rosow E, Mathew J, DeGraff AC. Non-Invasive Cardiopulmonary Function Assessment System, J Clin Monit 13, 413, 1997. 9. Johnson, GW, LabVIEW Graphical Programming: Practical Applications in Instrumentation and Control, 2nd ed., McGraw-Hill, New York, 1997. 10. Mathew JP, Adam J, Rosow E, Fontes ML, Davis L, Barash PG, Gillam L., Cardiovascular Pres- sure-Dimension Analysis System, J Clin Monit 13, 423, 1997. 11. National Instruments 1999 Measurement and Automation Catalog, National Instruments, Austin, TX. 12. Rosow, E. Technology and Sensors, Presented at the United States Olympic Committees Sports Equipment and Technology Conference, Colorado Springs, CO; November 19-23, 1992. 13. Rosow, E. Biomedical Applications using LabVIEW, Presented at the New England Society of Clinical Engineering, Sturbridge, MA; November 14, 1993. 14. Rosow, E., Adam, J.S., Nagle, J., Fisher, J.P. and Gillam, L.D.; Premise Development Corporation, Avon, CT and Hartford Hospital, Hartford, CT, A Cardiac Analysis System: LabVIEW and Acoustic Quantication (AQ), Presented at the Association for the Advancement of Medical Instrumenta- tion in Washington, D.C. May 21-25, 1994.