Digital Images PDF

# THE DIGITAL IMAGE FILE The basis of a digital image is a computer file that contains information relating to a signal that has been measured. In diagnostic medical imaging, the signal can be the pattern of x-ray emission from a patient (radiography), quantification of fan-beam x-ray attenuation b...

# THE DIGITAL IMAGE FILE The basis of a digital image is a computer file that contains information relating to a signal that has been measured. In diagnostic medical imaging, the signal can be the pattern of x-ray emission from a patient (radiography), quantification of fan-beam x-ray attenuation by the patient (computed tomography [CT]), the pattern of sound waves reflected by the patient (ultrasound), or the pattern of radiofrequency emission from the patient (magnetic resonance imaging [MRI]). In diagnostic medical imaging, the computer file is a Digital Imaging and Communications in Medicine (DICOM) format. A DICOM file is a computer file format of the same basic type as .jpeg and .tiff files, and is listed with a .dcm file extension. In addition to the actual image information, every DICOM file contains a large amount of ancillary information stored in what are called metadata tags. These tags contain other related information about the image, including the manufacturer of the device that generated the image; the date and time of image acquisition; patient demographic information; acquisition parameters; referrer, practitioner, and operator identifiers; and various other image parameters. ## THE COMPONENTS OF A DIGITAL IMAGE When the digital file is viewed using appropriate software, discussed later, the digital image comprises individual picture elements termed pixels. The pixels are arranged in a row-by-column matrix. In radiographic, ultrasound, CT, and MRIs, each pixel has a specified shade of gray. The more pixels in a digital file, the greater the matrix size and the larger the file size. Uncompressed digital radiographic image DICOM files are typically in the order of 4 to 12 MB per image, compared to approximately 0.1 MB for a 30-page document (.doc) file and 0.5 MB to 2.0 MB for a high-resolution, color, digital photographic (.jpg) file. The size of each pixel in a digital image determines the spatial resolution of the image, that is, how small of an object can be detected (Fig. 2.1). The hardware provided by the vendor determines the size of the pixel array in a digital radiographic image. In general, more pixels are better, but for any imaging system, there is a point at which more pixels do not result in a superior image. Pixel size (also termed pixel pitch) may be measured in microns or line pairs per mm; guidelines for minimum recommended standards for spatial resolution are available. Although pixel size is important, there are many other hardware and technical factors that ultimately affect the “clinically useful” spatial resolution. Also, detailed later, software manipulation of the raw image file to optimize the displayed image is, within limits, more important than pixel density. As noted before, in medical digital images, each pixel is assigned a shade of gray. The number of available gray shades for each pixel will influence the contrast of the image. If a pixel can have only a few gray shades, the image can only have a very short scale of contrast (high contrast) because there will be only black pixels, white pixels, and pixels of a few gray shades. If, conversely, there are many gray shade options for each pixel, the scale of contrast can be long. Computer files use binary notation to assign a gray shade to a pixel. In binary notation, the only digits used are 0 and 1. For grayscale assignment, a 0 is assigned black, and a 1 is assigned white; combinations of zeros and ones can be used for various intermediate gray shades. If n equals the number of allowable number of zeroes and ones per pixel, then 2^n is the number of combinations of zeroes and ones, and consequently the number of gray shades that are possible. The more combinations of zeroes and ones, the more gray shades there are possible in each pixel. The number of possible gray shades is referred to as the bit depth of the image (Fig. 2.2). Image file size and bit depth are directly related. Without the ability to assign multiple gray shades to each pixel, the image would have no diagnostic value (Fig. 2.3). ## DIGITAL RADIOGRAPHY ACQUISITION HARDWARE There are two main types of digital radiography acquisition hardware: computed radiography (CR) and direct digital radiography (DDR). As noted, a conventional x-ray tube and x-ray table are used for both CR and DDR applications. The difference between analog and digital imaging is that the film cassette, used for analog imaging, is replaced with a digital recording device to record the distribution of x-rays transmitted through the patient. In CR systems the digital recording device is a cassette that contains a flexible imaging plate, and in DDR systems the digital recording device is a rigid imaging plate or imaging chip (no cassette). Charge-coupled device (CCD) systems, which are a special type of DDR, may require purchase of a new x-ray table because the hardware components are housed in the x-ray table, and it may not be possible to retrofit the new equipment into an existing x-ray table adequately. ## Computed Radiography CR was the earliest digital radiographic system to come on the market. In CR, a cassette that appears outwardly similar to a film cassette is used. The CR cassette does not contain an intensifying screen or x-ray film but instead contains a flexible imaging plate coated with a photostimulable phosphor ([PSP]; Fig. 2.4). In CR, the x-ray attenuation pattern of the patient is temporarily stored as a latent image in the PSP. The latent image is created by changes in electron energy bands that result from x-rays striking the PSP. The latent image is read out optically as photostimulated luminescence when the PSP plate is subsequently stimulated using a scanning laser (Fig. 2.5). Therefore, the CR plate must be processed in a plate reader following radiographic exposure (see Fig. 2.4). The processing steps in the CR radiography process are summarized as follows: - After exposure, the CR cassette is removed from the x-ray table and inserted into the reader. - The reader automatically removes the PSP plate from the CR cassette. - The PSP plate is illuminated in the reader by a laser. - Laser illumination causes visible light to be emitted from the PSP plate. - Emitted visible light strikes a photomultiplier tube where it is converted into electronic signal. - The electronic signal is digitized and stored as a digital file. - The PSP plate is exposed to bright white light to make sure all electrons are at ground state in preparation for the next exposure. - The PSP plate is automatically returned to the cassette in the reader. - The cassette is ejected from the reader for use on the next patient. CR remains in widespread use in human imaging but is not as popular in veterinary imaging. There is nothing inherently wrong with CR imaging, and the reason for this relative lack of market penetration is multifactorial. Because the CR plate is handled and manipulated much the same way as an analog film cassette, it may take 1 to 2 minutes for the digital plate to be processed completely in the plate reader, whereas with DDR systems the image is available almost immediately. Because of this, CR does not increase workflow compared with analog imaging. For example, a CR plate reader that can read only one plate at a time is actually slower than a modern automatic x-ray film processor on a per-image basis. While CR cassettes are portable, the reader is typically not mobile and is permanently housed in the veterinary practice. In an ambulatory situation, where x-ray exposures are performed off-site (such as in a horse barn), exposed CR cassettes must be returned to the veterinary practice for processing, preventing rapid evaluation of radiographs in the field. This is a similar workflow as film radiography, so when throughput is not an issue, a good-quality CR system can be perfectly adequate as a replacement for an analog system. One advantage of a CR system is that damaged cassettes can be replaced relatively inexpensively, whereas the cost of replacing a damaged DDR plate, discussed in the next section, is considerably more expensive. Another advantage of CR is that the cassette is not tethered to a computer by a cable as nonwireless digital-imaging plates are. This provides more flexibility and reduces constraints in obtaining nonstandard views, such as images made with a horizontally directed x-ray beam, because the geometric relationship between the x-ray tube and CR plate can be adjusted easily. ## Direct Digital Radiography There are three distinct types of DDR: (1) indirect flat-panel detector system, (2) direct flat-panel detector system, and (3) CCD system. In DDR, the detector replaces the film cassette, and within a few seconds after the radiographic exposure, the digital radiographic image is ready for quality-control evaluation. ### Indirect Flat-Panel Detectors Indirect flat-panel detectors are termed indirect because they produce visible light as an intermediate step in image formation. An x-ray intensifying screen, identical in principle to those mentioned in Chapter 1, is used to convert x-ray energy emerging from the patient into visible light. The intensifying screen may be comprised of gadolinium oxysulfide or cesium iodide. Cesium iodide innately possess higher efficiency of x-ray detection compared to gadolinium oxysulfide, allowing lower radiation dose and thinner flat-panel device, but comes at a price premium. Regardless of composition, the intensifying screen is layered onto a panel containing an array of tiny photodiodes. Much as with the PSP found in the CR plate, the photodiodes convert visible light emitted from the intensifying screen into an electronic signal that is then read out by a thin-film transistor array and transformed into an electronic file (Fig. 2.6). To have good spatial resolution, the size of each detector element is very small, and a full-size imaging plate measuring approximately 43 cm × 43 cm may have a photodiode matrix of 2600 × 2600. Thus, an indirect flat-panel detector can have 6 to 7 million photodiodes or more. As can be imagined, the electronics needed to record and spatially localize the signal from each photodiode, or pixel, and to also incorporate the photodiode into each detector element is quite complicated. This contributes to the relatively high cost of flat-panel detectors compared with a film-screen cassette system. Indirect flat-panel detectors are commonly capable of a bit depth of 14, meaning there is a grayscale resolution of 16,384 gray shades per pixel. Most digital radiographic systems range in bit depth from 10 to 16 (1024-65,536 shades of gray). Because the human eye can register only 50 to 100 simultaneous shades of gray, even with extreme manipulation of image contrast (postprocessing-see later), there is probably little benefit in generating images with more than 14- to 16-bit depth. ### Direct Flat-Panel Detectors In a direct flat-panel detector, there is no visible light intermediary. Oncoming x-rays strike a photoconductor, typically composed of amorphous selenium, which has high efficiency x-ray absorption. Electrons liberated in the selenium layer by the oncoming x-ray beam are collected to form a charge. This charge is then read out directly by the thin-film transistor array, processed by the readout electronics and converted to an electronic file (see Fig. 2.6). As with indirect flat-panel detectors, there are millions of detectors in the array, and the pixel matrix is of comparable size, also with a bit depth of 14. Thus, the main difference between indirect and direct flat-panel detectors is the intermediate step of light production from the intensifying screen in indirect systems. As discussed in Chapter 1, there is some light diffusion within an x-ray intensifying screen. This light diffusion, which can potentially lead to blurring, is a purported disadvantage of indirect flat-panel detectors versus direct flat-panel detectors. However, with modern engineering techniques, and the structure of the crystals in the intensifying screen, the amount of light diffusion that actually occurs in the intensifying screen is minimal and not clinically significant. For all practical purposes, the functionality of modern-day indirect and direct flat-panel detectors is comparable. Flat-panel systems are increasingly replacing CR systems as the preferred digital radiograph technology in veterinary medicine, given their higher throughput and recent decreases in price. Flat-panel systems can produce diagnostic images with lower radiation doses (up to 30% less than CR systems), providing radiation safety benefit to personnel working in the x-ray suite . While direct and indirect flat-panel detectors are commonly tethered to a processing computer by a cable, wireless DDR plates are becoming more affordable and widely available (Fig. 2.7). Wireless DDR plates provide particular benefit in the equine setting, where additional cables are a tripping hazard. The recent advent of battery-powered, portable x-ray devices combined with a wireless flat-panel display provides the ultimate in portability and convenience for the equine clinician working in the field. Additionally, although CR is largely a fully developed technology, material improvements in flat-panel systems continue. Newer flat-panel detectors use a flexible plastic film as the base substrate for the thin-film transistor array, rather than the traditional silicone-based glass substrates currently in use, promising better durability and improved x-ray detection. Newer flat-panel detectors may also allow dynamic digital radiography, allowing rapid capture of multiple sequential images at once. ### Charged-Coupled Device CCD chips are used routinely in video camcorders and digital cameras, but their use in radiographic imaging is less common than either CR or flat-panel DDR radiography. However, some vendors continue to market veterinary radiography systems based on the CCD chip. CCD systems are included in the discussion of DDR because no equipment manipulation is needed between radiographic exposure and image visualization, such as is required in processing a CR cassette. CCD chips are relatively small compared with a flat-panel detector, being only a few centimeters on each side, compared with up to 43 cm for flat-panel detectors. Although small, there may be millions of pixel elements on the surface of a CCD chip. The CCD chip is sensitive to visible light, not x-rays; this is the basis for their use in camcorders and digital cameras. Thus, for radiography based on CCD technology, a light intermediary is necessary, and this is accomplished by integration of a relatively standard x-ray intensifying screen between the patient and CCD chip. When the CCD chip is exposed to visible light that has been collected and focused from the intensifying screen, the pixels in the CCD chip accumulate electronic charge, which is read out and converted to an electronic file. The geometry of the CCD system deserves special consideration. Given the small size of the CCD chip, the visible light output from the intensifying screen needs to be focused onto the chip, whereas in an indirect flat-panel detector, the intensifying screen is layered directly onto the detector. Focusing of light from the intensifying screen onto the CCD chip is usually accomplished by integration of fiberoptic light collection and a focusing lens between the intensifying screen and CCD chip (Fig. 2.8). Thus, the quality of the image depends more on the quality of light collection and focusing than on the quality of the CCD chip. Also, there is some loss of light and potentially much light distortion between the intensifying screen and CCD chip, leading to image degradation. In general, CCD radiography in humans has been reserved for situations where the part being imaged is relatively small (e.g., dental radiography and mammography) and there is less need for light focusing. In veterinary medicine, the parts being radiographed are larger and requirements for focusing more stringent; this can lead to poorer images than obtained with flat-panel detectors, and image quality from CCD systems is considered by many to be inferior to both flat-panel and CR systems. Also, the distance required for collecting and focusing the light from the intensifying screen means that the physical device is large compared with a flat-panel detector. As a result, CCD detectors are housed within the x-ray table. This hardware cannot be retrofitted easily and purchase of a new x-ray machine may be necessary if CCD technology is chosen. The size of the device also limits portability, and there is a requirement for a constant vertical relationship between the x-ray tube and CCD camera. Both of these factors eliminate the possibility of using CCD technology for portable imaging and horizontal beam radiography. CCD systems are typically less expensive than flat-panel imaging systems or high-quality CR systems, but image quality is usually inferior. A summary of the various hardware options is shown in Fig. 2.9.12.