Digital Radiography

When the first edition of the Journal of Thoracic Imaging was published in 1985, the radiology department was a very different kind of place from what it is today. It is remarkable to reflect on the evolution that has taken place so rapidly and completely, to the point where residents completing a radiology training program in 2009 may never have hung a film on a light box! The digital revolution has had a huge impact in just about every area of our existence, though surely none more than in the radiology department. In the radiology department of the 1980s, films were still exposed in cassettes containing fluorescent screens, or automatically transported from a dedicated chest unit to a chemical processor, to emerge fixed and dried after 90 seconds. It was a world of multiple color-coded film jackets, film bins, and grease pencils, where the film file room was the hub of the department. Batches of films were loaded on massive motorized film viewers (“multiviewers”) for reading throughout the day. “Hot lights” were placed strategically at every alternator to allow viewing of over exposed films. Entire teams of clinicians would descend on the radiology department each day to review their cases with the radiologists (there was only one set of films). These frequent encounters with visiting clinicians were arguably one of the few advantages of analog imaging, though in fact, many of these interruptions were redundant or inefficient. Inevitably, it was a system that became increasingly difficult to manage as the volume of procedures and the number of images per examination increased, together with the demands of multiple clinical services to have access to the same set of films. The more interesting the case and the more clinical services involved, the more likely it was that the films would be lost (when the filmless radiology department of the future was promised as a solution, a frustrated clinician caustically remarked that, in the case of his patients, we seemed to be already virtually “filmless”). As it happened, digital imaging arrived on the scene at a very opportune time. In fact, the promise of instant availability of any patient's images throughout the hospital was so irresistible that expectations became grossly inflated for a time, and frustration often followed. It was a standing joke for many years that a Picture Archiving and Communication System (PACS) was “always 5 years away” and then “always 2 years away” and so on, until it eventually became a reality. One tends to forget the magnitude of the obstacles that existed. As recently as 1994, when the University of Chicago was planning a PACS for a new outpatient facility, one of the major concerns was the cost of digital storage systems. Large robotic tape and optical disc archives were available. However, the lowest available cost was determined to be $1000 per gigabyte. Today, an external hard drive with one terabyte (1000 gigabytes) is available for less than $100 (10c per gigabyte). Workstations were also extremely expensive and bulky, consisting of arrays of massive cathode ray tube (CRT) monitors. A major manufacturer showing a prototype 4-monitor display at a meeting of the Radiological Society of North America in the early 1990s mentioned a tentative price point of $100,000 (for 1 workstation!). In the case of chest radiography, there was a special challenge because of the large and detailed nature of the image, and the wide range of gray scale required. Cathode CRT monitors were not up to the task of displaying chest radiographs with sufficient fidelity for rapid and accurate diagnosis.1,2 As a result, PACS was invariably implemented in stages. Softcopy reading was often used first in the intensive care unit (ICU) setting so that clinicians could enjoy the benefits of rapid availability of images at the point of care, whereas primary interpretation would be performed by radiologists using digital hardcopy mounted on a light box in the traditional way. In this way, many of the benefits of digital imaging were realized in terms of consistent image quality and the ability to reprint additional copies of images when needed. In contrast, use of film mandated batch reading with labor-intensive hanging and filing of films. In the case of bedside chest radiography in the ICUs, where the patients would often have multiple radiographs in one day, the films were usually maintained on dedicated multiviewers, with new films added and previous comparative films removed throughout the day as additional images were acquired. As comparison with previous images is essential in chest radiography, often with radiographs from several years earlier, there was a considerable period during which conventional screen-film radiographs were hung side-by-side for comparison with smaller digital hardcopy images. The hardcopy format that was initially employed was pioneered with the first commercial photostimulable phosphor radiography system in the United States in 1983, and employed a 50% scale image, with “conventional” and “processed” versions displayed side-by-side. The “processed” images had a wide effective latitude and relatively aggressive unsharp mask filtering to accentuate edges and lung details. This degree of miniaturization represented a compromise, and over time a larger single image hardcopy format approximating to 66% scale became established.3 During this transition period, there was much interest in film scanning as a way to achieve many of the benefits of digital imaging at a reasonable cost.4 At the University of Chicago, there was a period in the early nineties when all ICU portable chest radiographs were routinely scanned, and optimized by a “home-grown” image processing system, which applied automatic density and contrast correction with sharpening. Although conceived as a way to provide cost-effective 50% scale duplicates to the ICU clinicians, the results were so superior to the originals that the digital duplicates were soon used for primary interpretation, and the originals were filed and never viewed.5 In the mid and late eighties, numerous scientific papers were published that addressed the issues of image quality and diagnostic accuracy in digital chest radiography.6,7 Although the practical advantages of digital systems were obvious, there was concern that the lower spatial resolution that was available at that time might not be adequate for some clinical applications. Observer tests were performed, and comparisons were made using various pixel size displays and different display media. Hardcopy was compared with softcopy, and conventional gray scale was compared with reversed gray scale.8 Progressive improvements in softcopy displays and reductions in their cost eventually made the elimination of hardcopy a practical option. However, few of us realized at the time how long it would actually be before film would disappear from our reading rooms forever. Some departments were more aggressive than others in making the transition, and it is difficult to pinpoint the time at which the majority of departments in the United States had made the switch, but as recently as 2003, the chest reading area at the University of Chicago, one of the pioneer departments in digital imaging research, still used digital hardcopy for primary interpretation of chest radiographs. The rapidly falling cost of digital storage media and digital workstations actually provided an incentive to procrastinate in this regard. When softcopy reading was first implemented, the concept of manipulating the image during interpretation was new. It quickly became clear that the ability to extend the effective dynamic range by altering the window width and levels during interpretation was a great advantage, whereas the ability to magnify selected areas of the image was less useful, though still valuable in selected cases and for teaching. Although most of the focus was appropriately on the practical operational advantages of digital imaging during this time, the additional diagnostic potential was recognized at an early stage. For instance, scientific papers on computer-aided diagnosis for detection of lung nodules started to appear as early as the late 1980s, and the potential of dual energy imaging was also investigated.9–11 Dual energy did not become available commercially in the United States until 1996, at which time primary interpretation was still limited to hardcopy. Therefore, routine use of dual energy required printing additional images on film, or viewing the softcopy and bone images in selected cases on a CRT monitor using a separate application. For those of us who were already using energy subtraction routinely, the implementation of softcopy reading with the ability to flip electronically through a stack of standard, soft tissue and bone images by merely scrolling the mouse wheel was a big step forward. Computer-aided detection, in contrast, is only now starting to be used clinically for chest radiography. It is interesting to note how many years it has taken for these techniques to start to impact the mainstream of clinical work. Digital detector systems have also continued to evolve. During the mid-1990s, a unique selenium drum scanning system was introduced for chest radiography, with the promise of greater detection efficiency and higher image quality than was currently available with photostimulable phosphors.12 Shortly afterward, solid-state flat panel detector systems appeared on the market, though they were not adaptable to bedside imaging at that time.13 At the present time, both flat panel detectors and newer generation phosphor detectors compete in the market place, with the phosphor plates remaining dominant in bedside imaging, though solid state detectors have recently been adapted to bedside applications. In recent years, several investigators have addressed the potential of dose reduction in digital chest radiography, and particularly in younger patients, this has been a real benefit of digital systems.14,15 The ability to use a different effective “speed” on posteroanterior and lateral views with digital systems, with assignment of a proportionately greater dose to the posteroanterior projection, also provides opportunities for dose reduction, particularly in dual energy radiography. At the time of the original issue of the Journal of Thoracic Imaging in 1985, digital chest radiography was a relatively new and exciting modality. In the journal and elsewhere, there was no argument that digital imaging represented the future, though it was obviously not yet mature in many ways. An extensive review of digital radiography in that first edition provided an excellent introduction to the subject, with an explanation of basic concepts such as pixels, gray scale, and unsharp mask processing.16 Two articles in that same volume emphasized the virtues of selected screen-film applications (equalization radiography and screen-film tomography) over more “complex and expensive techniques such as digital radiography,” and indeed, at that time, the logic was unassailable.17,18 Nonetheless, the days of screen-film imaging were already numbered, and analog chest radiography was soon to be laid to rest, having served us well for almost exactly a hundred years. Now, as digital radiography has become established, we tend to take for granted the great increase in efficiency that has been achieved. Turnaround time from acquisition to final interpretation, which was once measured in days, now commonly occurs in a matter of minutes. Productivity, which was temporarily reduced with the initial transition to first-generation PACS, is far superior to what was achievable with film. Our ability to extract diagnostic information from the chest radiograph, by use of image manipulation, energy subtraction and potentially computer-aided diagnosis, is greater than ever. The promise of the fully digital radiology department has finally been realized, and for those of us who have lived through the transition, it has been truly revolutionary.