Digital Image Characteristics, Receptors, and Image Acquisition

Spatial Frequency and Spatial Resolution

Spatial resolution in digital imaging is primarily limited to the size of the pixel; however, when measuring an imaging system's ability to resolve small objects, it is important to understand the concept of spatial frequency and its relationship with spatial resolution. Spatial frequency can be defined by the unit of line pairs per millimeter (lp/mm). A resolution test pattern is a device used to record and measure line pairs (Fig. 4-6). Anatomic details are composed of large and small objects, and radiographic images display these details as variations from white to black brightness levels. Small objects have higher spatial frequency, and large objects have lower spatial frequency. It is more difficult to accurately image small anatomic objects (high spatial frequency) than to image large ones (low spatial frequency). An imaging system that can resolve a greater number of lp/mm (higher spatial frequency) has increased spatial resolution (Fig. 4-7). In digital imaging systems, the ability to resolve or demonstrate a specific spatial frequency is directly impacted by the size of the pixel. The images of the wrist (Fig. 4-8) demonstrate the impact that pixel size has on the spatial resolution visualized in an image.

Dynamic Range

The dynamic range of an imaging system refers to the ability of an IR to accurately capture the range of photon intensities that exit the patient. Digital IRs have a wide dynamic range (Fig. 4-9). In practical terms, this wide dynamic range means that a small degree of underexposure or overexposure would still result in diagnostic image quality. This characteristic of digital receptors is advantageous in situations where automatic exposure control (AEC) may not be available, such as in mobile radiography. However, radiographers must make considerable effort to select exposure techniques that produce diagnostic images without insufficient or excessive exposure to the IR. Although digital detectors have a wide dynamic range, exposure latitude refers to the range of exposures that should be used to produce a diagnostic image (see Fig. 4-9) and is likely established by the preferences or needs of the imaging department.

During digitization of the image, a numerical value (bit depth) is assigned to the pixel that represents an x-ray intensity based on the attenuation characteristics of that volume of tissue. A larger pixel bit depth means the image displayed has a wide grayscale, and this illustrates the wide dynamic range of the digital detector. As noted in Fig. 4-9, the digital IR can accurately capture more than 4000 x-ray intensities exiting the patient. To display that range of brightness levels (grayscale), the pixel bit depth is 12. If the digital IR is able to capture over 16,000 x-ray intensities exiting the patient, the bit depth would be 14, and contrast resolution is improved compared with the 12-bit digital system.

Processing the digital data yields a radiographic image that can be viewed on a display monitor and altered in various ways. Even if optimal exposure techniques are not used, the image rescaling that occurs during the processing stage can produce images with the appropriate brightness levels. Digital image processing and display are discussed in detail in Chapter 5.

The ability of the IR to capture a wide range of exit photon intensities does not mean a quality image is always created. Although lower-than-necessary x-ray exposures can be detected and processed, image quality suffers because there is insufficient exposure to the IR, and quantum noise results. The computer can process the data resulting from an IR exposed to higher-than-necessary radiation and produce a quality image but at the expense of patient overexposure. It is the responsibility of the radiographer to select the amount of exposure necessary to produce a diagnostic digital image (exposure latitude).

Dose Monitoring

Air kerma (kinetic energy released in matter) specifies the intensity of x-rays at a given point in air at a known distance from the focal spot or source of x-rays. DR systems use dose-area product (DAP) as an indicator of exposure. DAP is a measure of exposure in air, followed by computation to estimate absorbed dose to the patient. It is measured by a DAP meter embedded in the collimator. The DAP value depends on the exposure factors and field size and reflects both the dose to the patient and the total volume of tissue being irradiated. Kerma-area product (KAP) is the same as DAP and is the product of the total air kerma and the area of the x-ray beam at the entrance of the patient. Units of DAP or KAP are expressed in micro, milli, or centigray per area squared but can vary by manufacturer. DAP and KAP provide indicators of patient radiation risk and may be documented in the patient's record.

Modulation Transfer Function

As previously stated, a radiographic image displays a range of brightness levels (grayscale) based on the variation in radiation intensities exiting the tissue. Anatomic detail is best visualized when the brightness level of the object is different than its surrounding tissue (high contrast). Larger sized objects (low spatial frequency) are more easily visualized. As the size of the object decreases, it attains higher spatial frequency and becomes more difficult to visualize in a radiographic image. Modulation transfer function (MTF) is a measure of the imaging system's ability to display the contrast of anatomic objects varying in size, and the value will be between 0 (no difference in brightness levels) and 1.0 (maximum difference in brightness levels). The formula for MTF is as follows:

MTF=(Maximum intensity − Minimum intensity)(Maximum intensity + Minimum intensity)

An MTF of 1 (100% difference) would signify the difference between maximum and minimum brightness. An MTF of 1 is easier to achieve with large objects having low spatial frequency. It is more difficult to visualize smaller objects having high spatial frequency, and therefore most digital imaging systems' MTFs measure much lower than 1.0.

Detective Quantum Efficiency

Detective quantum efficiency (DQE) is a measurement of the efficiency of an IR in converting the x-ray exposure it receives to a quality radiographic image. If an IR system can convert x-ray exposure into a quality image with 100% efficiency (meaning no information loss), the DQE would measure 100% or 1.0. However, no imaging system has 100% conversion efficiency. Nevertheless, the higher the DQE of a system, the lower the radiation exposure required to produce a quality image, thereby decreasing patient exposure. The system's DQE value is impacted by the type of material used in the IR to capture the exit radiation (e.g., DQE is higher for DR compared with CR IRs) and the energy of the x-ray beam (e.g., DQE is higher for amorphous selenium coated [a-Se] at higher kilovoltage peak levels compared with cesium iodide [CsI]). The DQE is also impacted by spatial frequency and MTF. Some conversion efficiency is lost when imaging at higher spatial frequencies; therefore, DQE is directly proportional to the MTF of the detector.

Signal-to-Noise Ratio

Signal-to-noise ratio (SNR) is a method of describing the strength of the radiation exposure compared with the amount of noise apparent in a digital image. Noise is a concern with any electronic data set, in this case digital image noise. Because the photon intensities are converted to an electronic signal that is digitized by the ADC, the term signal refers to the strength or amount of radiation exposure captured by the IR to create the image. The varying x-ray intensities exiting the patient are converted to varying signal strengths. Increasing the SNR improves the quality of the digital image; this means that the strength of the signal is high in comparison with the amount of noise, and therefore, image quality is improved. Decreasing the SNR means that there is increased noise compared with the strength of the signal, and therefore, the quality of the radiographic image is degraded. Quantum noise results when there are too few x-ray photons captured by the IR to create a latent image. In addition to quantum noise, sources of noise include the electronics that capture, process, and display the digital image.

The ability to visualize anatomic tissues is affected by the SNR. Noise interferes with the signal strength just as background static would interfere with the clarity of music heard. When the digital image displays increased noise, regardless of the source, anatomic details have decreased visibility.

Contrast-to-Noise Ratio

Contrast-to-noise ratio (CNR) is a method of describing the contrast resolution compared with the amount of noise apparent in a digital image. Just as increased noise affects the SNR and visibility of the anatomic details, it also impacts the contrast displayed within the digital image. Brightness or signal differences in the digital image are a result of varying exit-radiation intensities from the attenuation of the x-ray beam in anatomic tissue (differential absorption). As previously stated, digital imaging systems have high contrast resolution. A system with higher contrast resolution means that anatomic tissues that attenuate the x-ray beam similarly (low subject contrast) can be better visualized. However, if the image has increased noise, the low subject contrast tissues will not be as well visualized. Digital images with a higher CNR increase the visibility of anatomic tissues (Fig. 4-10).

Different types of digital IRs use various methods of transforming the continuous exit radiation intensities into the array of discrete pixel values for image display. Some digital IRs use a sampling technique, whereas others have fixed detector elements (DELs) that are used to capture the exit radiation intensities. Regardless of the type used, a major determinant of the spatial resolution of digital images is the pixel size and spacing.

Computed Radiography

CR IRs come in many sizes and can be portable and used in a table or upright x-ray unit or fixed in a dedicated chest unit. The CR IR includes a cassette that houses the imaging plate (IP) (Fig. 4-11). The radiation exiting the patient interacts with the IP, where the photon intensities are absorbed by the phosphor layer of the IP. Although some of the absorbed energy is released as visible light (luminescence), a sufficient amount of energy is stored in the phosphor to produce a latent image. Luminescence is the emission of light when stimulated by radiation.

The IP primarily consists of support, phosphor, and protective layers (Fig. 4-12). The phosphor layer is composed of barium fluorohalide crystals doped with europium, referred to as the photostimulable phosphor (PSP). This type of phosphor emits visible light when stimulated by a high-intensity laser beam, a phenomenon termed photostimulable luminescence.

The phosphor (active layer) may be either a turbid or structured phosphor layer. A turbid phosphor has a random distribution of phosphor crystals within the active layer and can be used with both CR and DR IRs (Fig. 4-13). A structured phosphor layer has columnar phosphor crystals within the active layer resembling needles standing on end and packed together (Fig. 4-14). The reflective layer reflects light released during the reading phase toward the photodetector. The conductive layer reduces and conducts away static electricity. The support layer is a sturdy material to give some rigidity to the plate. Finally, the soft backing layer protects the back of the plate and assists in preventing backscatter (x-rays scattered back from the plate) from fogging the phosphor layer during exposure.

CR imaging requires a two-step process for image acquisition: image capture in the IP and image readout. The latent image is formed in the PSP when the exit x-ray intensities are absorbed by the phosphor and the europium atoms become ionized by the photoelectric effect. The absorbed energy excites the electrons, elevating them to a higher energy state, where they become stored or trapped in the conduction band (Fig. 4-15). The conduction band is an energy level just beyond the valence band (outermost energy band of an atom). The number and distribution of these trapped electrons (which form the latent image) are proportional to the exit exposure intensity as a result of the tissue's differential x-ray absorption. Some of these excited electrons immediately return to their normal state, and the excess energy is released as visible light. A percentage of electrons remain in this higher-energy state until released during laser beam scanning of the readout stage. The acquired image data (released energy) are extracted from the digital receptor, converted to digital data, and computer processed for image display. Exposed IPs should be processed within a relatively short amount of time (within 1 h) because the latent image dissipates over time (CR fading). CR fading occurs because some of the signal (released energy) captured in the IP is lost.

The exposed IP is placed in or sent to a reader unit that converts the analog data into digital data for computer processing (Fig. 4-16). Reader units are available in single- or multiplate configurations. The major components of a typical reader unit are a drive mechanism to move the IP through the scanning process; an optical system, which includes the laser, beam-shaping optics, collecting optics, and optical filters; a photodetector, such as a photomultiplier tube (PMT); and an ADC. Manufacturers differ in the CR reader mechanics. Some devices move the IP, and some move the optical components. There are three important stages in digitizing the CR latent image: scanning, sampling, and quantization.

The purpose of scanning is to convert the latent image (released energy) into an electrical signal (voltage) that can be subsequently digitized and displayed as a manifest digital image. Once in the reader unit, the IP is removed from the cassette and scanned with a helium–neon laser beam or a solid-state laser diode to release the stored energy as visible light (Fig. 4-17). Absorption of the laser beam energy releases the trapped electrons, and they return to a lower energy state. During this process, the excess energy is emitted as visible light (photostimulable luminescence). The scanning of the plate results in a continuous pattern of light intensities being sent to the PMT or photodetector, whose output is directed to the ADC for sampling and quantization.

A photodetector collects, amplifies, and converts the visible light to an electrical signal proportional to the range of energies stored in the IP. The collected electronic signal values are of low voltage and therefore need to be amplified by the photodetector in order to travel to the ADC. The signal output from the photodetector is digitized by an ADC in order to produce a digital image. To digitize the analog signal from the photodetector, it must first be sampled. An important performance characteristic of an ADC is the sampling frequency, which determines how often the analog signal is reproduced in its discrete digitized form (Fig. 4-18).

As mentioned previously, small anatomic details have higher spatial frequency and would therefore need a higher sampling frequency than low spatial frequency anatomic details. The Nyquist frequency is a standard formula for converting analog data into to discreet digital units to accurately represent the analog signal or imaging data in digital radiography. To accurately reproduce an image from the continuous analog signal, the sampling rate must be at least two times the highest spatial frequency in the exit x-ray intensities (signal). If the sampling frequency is too low, an improper waveform (aliasing) may result, which is considered an image artifact and reduces the visibility of small anatomic detail (Fig. 4-19).

Increasing the sampling frequency of the analog signal increases the pixel density of the digital data and improves the spatial resolution of the digital image. The closer the samples are to each other (increased sampling frequency), the smaller the sampling pitch, or distance between the sampling points (Fig. 4-20). Increasing the sampling frequency decreases the sampling pitch and results in smaller-sized pixels. The distance between the midpoint of one pixel to the midpoint of an adjacent pixel describes the pixel pitch. Spatial resolution is improved with an increased number of smaller pixels, resulting in a more faithful digital representation of the acquired analog image.

Manufacturers of CR equipment vary in the method of sampling IPs of different sizes. Some manufacturers fix the sampling frequency to maintain a fixed spatial resolution, whereas others vary the sampling frequency to maintain a fixed matrix size. If the spatial resolution is fixed, the image matrix size is simply proportional to the IP size. A larger IP has a larger matrix to maintain spatial resolution (Fig. 4-21). If the matrix size is fixed, changing the size of the IP would affect the spatial resolution of the digital image. For example, under a fixed matrix size system, changing from a 14 × 17 inch (35 × 43 cm) IR size to a 10 × 12 inch (25 × 30 cm) IR would result in improved spatial resolution for the same FOV (Fig. 4-22). Spatial resolution is improved because in order to maintain the same matrix size and number of pixels, the pixels must be smaller in size. It is recommended to use the smallest IR size reasonable for the anatomic area of interest.

Another important ADC performance characteristic is degree of quantization or pixel bit depth, which controls the number of gray shades or contrast resolution of the image. During the process of quantization, each pixel, representing a brightness value, is assigned a numerical value. Quantization reflects the precision with which each sampled point is recorded. As previously discussed, the pixel size and pitch determine the spatial resolution, and the pixel bit depth determines the system's ability to display a range of shades of gray to represent anatomic tissues. Pixel bit depth is fixed by the choice of ADC, and CR systems manufactured with a greater pixel bit depth (i.e., 16-bit [in which 2¹⁶ bits can display 65,536 shades of gray]) improve the contrast resolution of the digital image.

Before the IP is returned to service, the plate is exposed to an intense white light to release any residual energy that could affect future exposures. PSPs can be reused and are estimated to have a life of 10,000 readings before requiring replacement. Advancements in PSP material, laser beam technology, and dual-sided IP scanning will continue to improve the process of CR image acquisition.

Direct Radiography

DR IRs have a self-scanning readout mechanism that uses an array of x-ray detectors that receive the exit radiation and convert the varying x-ray intensities into proportional electronic signals for digitization (Fig. 4-23). In contrast to CR, which requires a two-step image acquisition process and results in a longer delay between image capture and image readout, DR imaging combines the two processes. As a result, DR images are available almost instantly after exposure. However, DR receptors are more fragile and much more expensive than CR IRs. Several types of electronic detectors are available for DR.