| | Digital Echocardiography: The Guiding Role of Harvey FeigenbaumEchocardiography has enjoyed remarkable technical progress and ever-expanding clinical application during the past half-century: M-mode in the 1950s; B-mode and spectral Doppler applications in the 1970s; color Doppler, transesophageal, and stress applications in the 1980s; tissue Doppler and three-dimensional (3D) applications in the 1990s; and strain imaging and real-time 3D in the new century. Arguably, one of the fundamental enabling technologies that has most advanced both clinical use and technical progress in echocardiography has been the evolution of all-digital storage, retrieval, and viewing during the past two decades. Harvey Feigenbaum was one of the earliest thought leaders to recognize the critical advantage that digital storage of two-dimensional (2D) echocardiograms would give, allowing random access to current and prior studies from anywhere in a hospital, or, for that matter, anywhere in the world. In this article, I will tie together the technical background and historical development of digital echocardiography, focusing on the seminal role that Harvey Feigenbaum has played. Ultrasound evolved as an analog technique, with acoustic signals being amplified and displayed on an oscilloscope or recorded (for M-mode) onto Polaroid film (Polaroid, Waltham, MA) or strip chart paper. Reams of strip chart paper were prominent in echocardiography laboratories of the 1970s and early 1980s, unfurled on the floor for photometric development and constituting a major expense for echocardiography laboratories. Digitization of M-modes was virtually unknown, except for research applications of endocardial borders hand traced using computerized tablets. B-mode ultrasound initially presented even greater problems, because early ones literally drew individual scan lines on an oscilloscope,1, 2 and the principal way to record the images was with a movie camera. One of the most important early applications of digital (computerized) technology came in the late 1970s, with the development of digital scan conversion.3, 4 For the first time, the polar data set that originated in a 2D echocardiograph could be displayed in a raster format, allowing it to be recorded on videotape. It also meant that at certain points within the processing stream of the machine, the image existed in a purely digital form and potentially could be stored in that format. At first, the ability to record 2D echocardiograms on videotape seemed to be the ideal archival solution. Once the beta-VHS wars were settled in favor of VHS, there was near-universal adoption of that videotape format in echocardiography laboratories. They offered a compact storage method for recording an entire echocardiography examination, allowing virtually all the information from the initial examination to be reviewed by over-readers and clinicians, albeit with somewhat degraded image quality. Videotape also was a very cost-effective medium, being leveraged off the mass market of the entertainment industry (an industry representative once remarked to me that, had the VCR been developed solely for the medical industry, the cheapest unit would have cost more than $10,000). Costs for videotape decreased as low as 25 cents/examination. However, despite these clear advantages, videotape storage suffered significant limitations, and Harvey Feigenbaum was among the first to point these out5: difficulty finding a given examination on a videotape, the need for multiple rewinding and reviewing of complex sequences (compounded by the painfully small knobs for frame-by-frame advance on some of the early VCRs), limited options for quantification, image degradation with each videotape duplication and indeed with the simple passage of time, and the complexity of comparing serial studies (so inconvenient that it was very rarely done). The initial impetus for digital storage and review of echocardiograms originated in Dr Feigenbaum’s laboratory in the late 1970s as a way to make computer-assisted measurements by overlaying quantitative electronic calipers over a frozen echocardiographic image. Working with John Freeland and Roger Camp, they used a Sony videodisk (Sony Corporation, Tokyo, Japan) to record 10 seconds of video to present a crisp frozen image to the reviewer, without the jitter of videotape machines of that era. To use this technology, they programmed software to overlay a measurement package over the image to run on a Zilog z80 microprocessor (ZILOG, San Jose, CA). Figure 1 shows the initial commercial offering from this effort, the Easy View. Shown below the monitor are the individual plug-in modules (512 KB of assembly-language code) for guiding measurements in 2D, Doppler, and obstetric ultrasound among others. It was with this technology that the first digital stress echocardiography was performed in 1981.6 When Sony discontinued the videodisk, they turned to the embryonic technology of computer frame grabbers, capable of capturing the image from the video port of the echocardiography machine (or secondarily from videotape) and storing it in digital format into 2MB of dynamic RAM in an early IBM personal computer (IBM, Armonk, NY). This technology proved to be instrumental in a number of the quantitative clinical and experimental studies coming out of the laboratory in that time period, involving among many others acute myocardial infarction, coronary anatomy, endocarditis, pericardial effusion, and right heart disease.7, 8, 9, 10, 11, 12 Although echocardiographic frame grabbing was developed simply to facilitate accurate measurements, often without even archiving the digitized images once measurements were made, Dr Feigenbaum soon recognized other advantages of digital echocardiographic storage and review, including easier interpretation, more convenient sharing with clinicians and at educational conferences, facilitated comparison with prior studies, and rendering practical the nascent technique of stress echocardiography. Recognizing these fundamental advantages of digital echocardiography and, in particular, devising technical implementations within the severe limitations of computer technology at the time remain Dr Feigenbaum’s enduring legacy in digital echocardiography. To understand the challenge that Dr Feigenbaum and his engineering collaborators faced in implementing this approach, it is important to recall the demands of full digital echocardiography versus the primitive capabilities of computer hardware in the late 1970s and early 1980s. In comparison with nuclear cardiology (which was also undergoing a transition to digital storage about the same time), echocardiography requires orders of magnitude more storage. For example, a typical videotape echocardiographic study lasts 10 minutes at 30 frames/s (a total of 18,000 frames). Each frame would ideally be digitized at about 512 rows × 512 columns (approximately 250K pixels), each requiring 24 bits (3 bytes) for color storage. Multiplying this out yields more than 14 gigabytes of storage needed for a single 10-minute clinical study. Contrast this with the embryonic storage technology of the time, which on the initial IBM personal computer consisted of 360-KB floppy disks, almost 40,000 of which would be needed to store that typical echocardiographic study! Rather than dismay and settle for the limitations of videotape, Dr Feigenbaum and colleagues sought any way that could reduce storage requirements. First, they reasoned that it was necessary to store only one cardiac cycle per view, because it could be played repeatedly for careful analysis. Assuming about 30 seconds of recording for each view, this clinical compression instantly reduced storage needs 30-fold. Next, it was recognized that although 512 × 512 digitization would be nice, 256 × 256 storage would capture most of the clinically relevant data, saving another 75% of storage. Third, they reasoned that most of the action was in systole (a fact that even this diastologist would have to accept) and stored only 8 frames of the typical 30 in a cardiac cycle, another 73% saving, capturing them at 20 frames/s but playing back at 10 frames/s to restore the proper tempo of the heartbeat. Fourth, they were willing to sacrifice color and dynamic range, initially storing only 16 shades of gray (4 bits of data). These measures in total yielded 2700:1 compression, eliminating more than 99.96% of data per view. Without any digital compression, this requires about 260 KB per loop, storable on the early floppy disks. With modest 10:1 digital compression, this allowed all the standard views of the echocardiogram to be stored on a 360-KB floppy, and thus, in 1984, the era of digital echocardiography was begun. Dr Feigenbaum clearly recognized that such truncated storage did not in any way represent a complete echocardiographic study5: “The important message is that limitations and compromises are a part of digital recordings. One must constantly keep in mind the purpose of recording the images in this fashion. One should not necessarily attempt to digitize everything…on the real-time examination. Digital recordings are not intended to replace videotape…[and] are primarily for convenience, rapid review, serial studies, and quantitation.” The videotape recording was always reviewed in detail as part of the official interpretation. He also knew that computer evolution was on our side: “It is possible that as digital technology improves and as we are able to store more information faster and at less cost, some of the compromises may be lessened.” Clinical Applications  Once digital storage was available within the Indiana University laboratory (Indianapolis, IN), it contributed to virtually all the research conducted within the laboratory. For example, the developing field of stress echocardiography was greatly boosted by the ability to view pre-exercise and postexercise images side-by-side.13 The results of aneurysmectomy could be directly compared with preoperative images.14 Even assessment of coronary artery anatomy could be facilitated by looping images to remove respirophasic movement.15 Furthermore, the availability of image data in digital format allowed averaging to improve measurement accuracy16 and color coding of endocardial movement to facilitate interpretation.17 Evolution of Digital Echocardiography Although Dr Feigenbaum used digitizing equipment from numerous vendors, including Micro Sonics (Indianapolis, IN), Nova Microsonics (Allendale, NJ), and Tomtec (TomTec Imaging Systems, Munich, Germany), the overall architecture of his digital echocardiography laboratory was largely a homegrown one, programmed in part by Jeff Fitch and maintained by Dave Wagoner. The initial storage medium was 5.25-in floppy disks, and storing one or more disks per patient rapidly lead to every available wall in the laboratory being lined ceiling to floor with shelves of these floppies (Figure 2). When the computer world evolved to the smaller and higher capacity 3.5-in floppies, Dr Feigenbaum initially resisted (perhaps not wishing to remodel all those shelves!), but he ultimately embraced this change with passion and later the move to still higher capacity CD-ROMs, magneto-optical disks, digital jukeboxes, and networked systems. His eagerness to surf the technologic wave embodied in Moore’s law12 helped greatly to push digital echocardiography into the clinical realm. Figure 3 shows a later digital acquisition system, ca. 1991 to 1993. By the early 1990s, computer technology had evolved to the point that digital echocardiography was much more feasible. Importantly, a number of the major echocardiography manufacturers began developing their own digital output options. This allowed echocardiographic images to be stored from the internal digital data, rather than requiring digitization of the video output, which always had some degradation, although not as much as digitized videotape. Unfortunately, there was no agreement among manufacturers as to the format with which to store the data, and thus a reading station from one company could not read echocardiographic data from any other company. Although single-vendor laboratories would not be greatly impacted by this lack of interoperability, this emerging tower of Babel threatened to limit exchange of echocardiograms between laboratories, critically impacting telemedicine with echocardiography. To overcome these incompatible formats, in 1993 the American Society of Echocardiography (ASE) joined with engineers from all major vendors to develop an echocardiographic version of the digital imaging and communication in medicine (DICOM) standard.18, 19 I served as the ASE representative to the DICOM committee during this time period, and during the next couple of years, we met roughly once a month to hammer out the technical details of this standard, allowing the exchange of echocardiographic studies by network and on fixed media, such as CDs. It was a remarkable process, where vendors who were otherwise fierce competitors would come together to find common ground and devise the best standard for the field. There were many issues to resolve: store raster images (rectangular, like television or a computer monitor) or polar (sector) data, derived from the original scan line data? (Raster won.) Store the pixel data as separate layers of B-mode and color Doppler data (as Hewlett-Packard [Philips, Andover, MA] did at the time) or as a single layer of red-green-blue data (as Acuson [Siemens, Mountain View, CA] did)? (In a typical compromise, both were allowed; to be DICOM compatible, a display workstation had to display both, along with a number of other variations.) There were many debates regarding the number of rows and columns to store and other resolution issues. I remember well my many discussions with Dr Feigenbaum during this time, seeking his advice. He was quite clear: do not make it so complex or technically demanding that it will keep people from implementing digital echocardiography. As the DICOM standard was formally ratified around 1995, the major echocardiography vendors began implementing digital echocardiographic archiving and display solutions, along with several of the traditional PACS vendors from radiology. In general, Dr Feigenbaum resisted digital echocardiographic systems from the major ultrasound instrument manufacturers, reasoning that they could not help but show their own images just a little better than their competitors’ images. He held on to his homegrown system as long as possible, but by the late 1990s he had hit a brick wall in the form of the 64-KB memory segmentation of his MS-DOS–based system, which could not easily display the larger images and loops now being produced. He needed to migrate to Windows, but his programmers at the time had no experience. Enter his son, Tom, a physics graduate from Duke (Durham, NC) with a master’s degree from the University of Illinois (Urbana-Champaign, IL). Some years before he had founded Problem Solving Concepts, a software development company aimed at the engineering community. In 1998, Harvey proposed to Tom that he migrate his digital echocardiography laboratory to Windows (Microsoft Corporation, Redmond, WA). As it turned out, Problem Solving Concepts’ development environment was ideally suited to this type of application, and within a month in 1998, starting from scratch, they had a working prototype for the laboratory. According to Tom, his father played a pivotal role in the development of the system, insisting on several things: it had to display DICOM images from all vendors with equal fidelity; all prior echocardiograms had to be available for side-by-side review during study interpretation; and image measurement had to be quick and intuitive. Just as important, it had to be software only, able to run on commercial off-the-shelf computers, thus keeping the cost down and the speed high. Although initially developed solely to modernize the Indiana University laboratory, Dr Feigenbaum soon recognized that the program could help other laboratories move to digital echocardiography and encouraged Tom to commercialize the project. After a year of due diligence, ProSolv Cardiovascular was marketed to the echocardiographic community. In an environment with strong offerings from the major instrument and PACS manufacturers, ProSolv found success by adhering to Dr Feigenbaum’s principles: fast (using off-the-shelf hardware), cost-effective (software only model), nonsectarian (everyone’s images shown equally well), and with a very clean and logical user interface that one observer noted, “must have been designed by an echocardiographer,” as indeed it was! I am sure many readers have heard Dr Feigenbaum’s lecture, “The ABC’s of Digital Echo”: A is for accuracy, because digital echocardiography should improve diagnosis; B is for business, because digital echocardiography should make laboratories more efficient; and C is for communication, because digital echocardiography should help disseminate study results and images to referring physicians. So where is digital echocardiography in 2008? The advantages of digital acquisition, storage, and analysis are largely taken as a given, and the majority of major laboratories have made the transition to predominant digital imaging. The ASE has strongly encouraged this movement.20 Computers keep getting faster and faster with cheaper mass storage, which is fortunate because our digital echocardiographic studies keep getting larger and larger, because of a combination of higher resolution imaging (commonly 800 × 600 vs the former 640 × 480 pixels), higher frame rates (80 to as much as 150 Hz, rather than the prior 30 Hz that was tied to the television standard), 3D imaging, and the storage of raw data formats in addition to the DICOM data. The DICOM standard itself has proven slow to adapt to this changing ultrasound technology, largely because of a deliberate desire for stability to ensure interoperability and backward compatibility of all implementations. Three-dimensional imaging is a good example of this slow approach. In the original DICOM ultrasound information object definition from the mid-1990s, 3D echocardiography was included in a very crude way, allowing a 2D image to be related to an external frame of reference by a series of translations and rotations. Almost immediately an effort was undertaken to develop a more comprehensive multidimensional image object, allowing non-Cartesian data organization (eg, based on the original scan line orientation rather than a rectangular representation) and storage of nonimage data (eg, raw radiofrequency or Doppler shift data). Unfortunately, after many years of work on this, the parent DICOM gatekeeper, Working Group 6, vetoed the proposal, stating it was too different from 3D representations for any other modality and thus would be impractical to implement in practice. An intensive effort to develop an acceptable 3D information object definition began in 2006 and is now released for public comment before voting on its inclusion in the standard. What this new representation gives up in the nuances of ultrasound acquisition, it should reclaim by having a standard format that can be more easily viewed and analyzed by visualization algorithms developed for computed tomography and magnetic resonance imaging. This affirms a strong push within the standards community toward multimodality imaging, moving away from modality-specific workstations. Leveraging advances in digital echocardiography from other industries has long been a theme of Dr Feigenbaum’s, and there is no question that echocardiography has been the beneficiary of investments in entertainment, communications, and computer science. Other current efforts in the digital imaging community include structured reporting, DICOM work lists, and Integrating the Healthcare Enterprise. Structured (results) reporting provides a standardized way to communicate the quantitative and diagnostic results from a study. Working Group 12 has recently developed the specific codes and measurements necessary to report adult echocardiography, whereas a concerted effort to port this to the pediatric realm is in its final stages. DICOM work lists address one of the greatest sources of errors in digital echocardiography, mistakes in entering patient names and hospital number, which may render a study impossible to find subsequently. By providing a link between hospital registration information (encoded in HL-7) and DICOM imaging software, all patient information can automatically be transferred to the echocardiographic machine. Integrating the Healthcare Enterprise21 is an industry-wide collaboration to go beyond formatting standards to develop practical procedures to integrate imaging throughout a patient’s health care experience. Conclusions The field of digital echocardiography has shown dramatic progress in the past 30 years, thanks in large part to the 1,000,000-fold increase in computer power that has occurred in that time. Equally important, however, has been the vision of true pioneers, who recognized the need for digital storage and divined ways of achieving their vision within the limitations of early computer systems. Foremost among these pioneers is Harvey Feigenbaum, who has maintained clarity of focus in digital echocardiography from the beginning to make it the universally embraced technology that it is today. The author gratefully acknowledges the memories contributed by Dr Feigenbaum’s colleagues and collaborators, including John Freeland, Aaron Waitz, Tom Ryan, MD, Arthur E. Weyman, MD, and Dr Feigenbaum’s son, Tom. References 1. 1Bom N, Lancee CT, van Zwieten G, Kloster FE, Roelandt J. Multiscan echocardiography, I: technical description. Circulation. 1973;48:1066–1074. MEDLINE 2. 2Griffith JM, Henry WL. A sector scanner for real time two-dimensional echocardiography. Circulation. 1974;49:1147–1152. MEDLINE 3. 3Ophir J, Maklad NF. Digital scan converters in diagnostic ultrasound imaging. Proc IEEE. 1979;654:654–664. 4. 4Leavitt S, Hunt B, Larsen H. A scan conversion algorithm for displaying ultrasound images. Hewlett-Packard J. 1983;34:30–34. 5. 5Feigenbaum H. Digital recording, display, and storage of echocardiograms. J Am Soc Echocardiogr. 1988;1:378–383. MEDLINE 6. 6Robertson WS, Feigenbaum H, Armstrong WF, Dillon JC, O’Donnell J, McHenry PW. Exercise echocardiography: a clinically practical addition in the evaluation of coronary artery disease. J Am Coll Cardiol. 1983;2:1085–1091. MEDLINE 7. 7Stafford A, Wann LS, Dillon JC, Weyman AE, Feigenbaum H. Serial echocardiographic appearance of healing bacterial vegetations. Am J Cardiol. 1979;44:754–760. MEDLINE |
CrossRef
8. 8Bauman W, Wann LS, Childress R, Weyman AE, Feigenbaum H, Dillon J. Mid systolic notching of the pulmonary valve in the absence of pulmonary hypertension. Am J Cardiol. 1979;43:1049–1052. MEDLINE |
CrossRef
9. 9Rogers EW, Feigenbaum H, Weyman AE, Godley RW, Vakili ST. Evaluation of left coronary artery anatomy in vitro by cross-sectional echocardiography. Circulation. 1980;62:782–787. MEDLINE 10. 10Heger JJ, Weyman AE, Wann LS, Rogers EW, Dillon JC, Feigenbaum H. Cross-sectional echocardiographic analysis of the extent of left ventricular asynergy in acute myocardial infarction. Circulation. 1980;61:1113–1118. MEDLINE 11. 11Armstrong WF, Schilt BF, Helper DJ, Dillon JC, Feigenbaum H. Diastolic collapse of the right ventricle with cardiac tamponade: an echocardiographic study. Circulation. 1982;65:1491–1496. MEDLINE 12. 12Moore GE. Cramming more components onto integrated circuits. Electronics. 1965;38:114–117. 13. 13Armstrong WF, O’Donnell J, Dillon JC, McHenry PL, Morris SN, Feigenbaum H. Complementary value of two-dimensional exercise echocardiography to routine treadmill exercise testing. Ann Intern Med. 1986;105:829–835. MEDLINE 14. 14Ryan T, Petrovic O, Armstrong WF, Dillon JC, Feigenbaum H. Quantitative two-dimensional echocardiographic assessment of patients undergoing left ventricular aneurysmectomy. Am Heart J. 1986;111:714–720. MEDLINE |
CrossRef
15. 15Presti CF, Feigenbaum H, Armstrong WF, Ryan T, Dillon JC. Digital two-dimensional echocardiographic imaging of the proximal left anterior descending coronary artery. Am J Cardiol. 1987;60:1254–1259. MEDLINE |
CrossRef
16. 16Petrovic O, Feigenbaum H, Armstrong WF, Ryan T, West SR, Green-Hess D, et al. Digital averaging to facilitate two-dimensional echocardiographic measurements. J Clin Ultrasound. 1986;14:367–372. MEDLINE |
CrossRef
17. 17Bates JR, Ryan T, Rimmerman CM, Segar DS, Sawada SG, Fitch G, et al. Color coding of digitized echocardiograms: description of a new technique and application in detecting and correcting for cardiac translation. J Am Soc Echocardiogr. 1994;7:363–369. MEDLINE 18. 18Thomas JD. The DICOM image formatting standard: what it means for echocardiographers. J Am Soc Echocardiogr. 1995;8:319–327.
Full-Text PDF (2075 KB)
|
CrossRef
19. 19Thomas JD, Khandheria BK. Digital formatting standards in medical imaging: a primer for echocardiographers. J Am Soc Echocardiogr. 1994;7:100–104. MEDLINE 20. 20Thomas JD, Adams DB, Devries S, Ehler D, Greenberg N, Garcia M, et al. Guidelines and recommendations for digital echocardiography. J Am Soc Echocardiogr. 2005;18:287–297. Full Text |
Full-Text PDF (128 KB)
|
CrossRef
21. 21Vegoda P. Introducing the IHE (integrating the healthcare enterprise) concept. J Health Inf Manag. 2002;16:22–24. Department of Cardiovascular Medicine, Cleveland Clinic Foundation, Cleveland, Ohio.  Reprint requests: James D. Thomas, MD, Department of Cardiovascular Medicine, Cleveland Clinic Foundation, 9500 Euclid Ave, Desk F-15, Cleveland, OH 44195.
PII: S0894-7317(07)00807-3 doi:10.1016/j.echo.2007.11.002 © 2008 American Society of Echocardiography. Published by Elsevier Inc. All rights reserved. | |
|
FULL TEXT00807-3/journalimage?img=PIIS0894731707008073.gr1.lrg.gif&fig=fig1&kwhquery=&issn=0894-7317&ishighres=false&allhighres=false&free=yes','journalimage');)
00807-3/journalimage?img=PIIS0894731707008073.gr2.lrg.gif&fig=fig2&kwhquery=&issn=0894-7317&ishighres=false&allhighres=false&free=yes','journalimage');)
00807-3/journalimage?img=PIIS0894731707008073.gr3.lrg.gif&fig=fig3&kwhquery=&issn=0894-7317&ishighres=false&allhighres=false&free=yes','journalimage');)