CH 20 Digital Image Processing

• ADC (analog-to-digital conversion), bit, byte, computed radiography (CR), detective quantum efficiency (DQE), deviation index (DI), digital radiography (DR), direct conversion, grayscale bit depth, high-pass filtering, histogram, indirect conversion, look-up table (LUT), low-pass filtering, kernel, matrix, photostimulable storage phosphor imaging plate, pixel, quantification, rescaling, signal-to-noise ratio, spatial resolution, voxel, and windowing are key terms in digital image processing.

Objectives of Digital Image Processing

• Discuss the types of digital radiography (DR) imaging systems
• Describe the digital image data formation process
• Explain digital image processing operations
• Describe the process of histogram acqusition and look-up table (LUT) application
• Explain the function of digital image window level and width controls
• Describe the factors affecting digital image quality
• Explain how exposure indicators can be a quality control tool for image quality and radiation protection

Historical Development

• Radiology transitioned to the digital world with the computed tomography discovery (CT) in the 1970s.
• Between the 1970s and 1990s, imaging sciences exploded due to digital computerization of CT and diagnostic medical sonography
• Digital processing is now applied to MRI, nuclear medicine, cardiovascular imaging, and mammography
• Digital radiography imaging systems have replaced traditional film with reusable detectors:
  • Computed radiography (CR)
  • Digital radiography (DR)
• CR systems use a photostimulable storage phosphor imaging plate (PSP or IP) inside a cassette
• The cassette is taken to a reader to process the plate and create the image after exposure
• CR systems involve a two-step process where the radiographer moves the detector between image acquisition and display
• Some PSP systems are permanently fixed in units that don't require this two-step process
• DR systems typically have the detector and reader as a permanent part of a table or wall unit
• Newer wireless DR systems resemble a sealed cassette that can be moved around a room
• DR systems acquire the image and send it directly to the display monitor
• DR systems can be categorized as using either:
  • Direct conversion
  • Indirect conversion
• Indirect conversion detectors use a scintillator to convert x-ray photons to light and a photodetector to convert light to an electronic signal
• CR (PSP technology), charge-coupled device (CCD), or amorphous silicon with a thin film transistor (TFT) array are examples of indirect systems
• A redefinition of basic imaging principles was brought about by the transformation of radiography from a film-based to digital imaging
• Changes in procedure reimbursements hastened the migration from film-based imaging to digital technologies
• Reimbursements for film-based x-ray procedures were cut by 20% starting in 2017
• CR-based procedures were reduced annually by 7% through 2018 and 10% thereafter
• DR procedures receive full reimbursements as incentives
• The Digital Imaging and Communication in Medicine (DICOM) is a standard to ensure that all equipment speaks the same computer language

Digital Image Formation

• Digital radiography imaging systems now use a reusable detector to produce a digital image instead of film
• Digital images are converted into numerical values for transmission or processing
• Radiographic images are produced using electronic devices
• Analog signals consist of voltage fluctuations on a waveform with infinite values
• Computers operate from a binary machine language
• Analog signals must be converted to a binary language through a process called ADC (analog-to-digital conversion)
• ADC consists of sampling and quantification
• Analog voltage values are measured at a chosen sampling frequency on the analog waveform during sampling
• Each sampled piece of analog data is then computed and assigned a discrete analog value through quantification
• Analog values are converted to a binary digit using binary counting methods
• The binary system uses two-symbol alphabet
• Electrical currents are either on or off, which makes the binary system consist of information recorded as either "0" or "1"
• "Bit" means binary digit
• The numbers 1 through 10 are written in binary code
• An 8-bit word is called a byte

Digital Image Characteristics

• Computerized digital images are described by the number of values displayed per image
• A matrix is a series of boxes in rows and columns that gives form to the image
• Image matrix boxes display numerical values that can be transformed into a visual brightness or gray levels
• Picture elements or pixels are the individual matrix boxes
• Each pixel location is determined by its address
• The total number of pixels in the matrix = the number of boxes in the row multiplied by the number of boxes in the column
• Each pixel represents a two-dimensional data point
• Each data point has a third dimension, creating a volume data point along the z-axis in other imaging modalities e.g. MRI
• A volume element or voxel is that three-dimensional data point
• The overall dimension of the image matrix is called the field of view (FOV)
• In digital radiography imaging systems, the FOV is determined by the detector size
• Pixels get smaller and images become sharper as matrix size increases if the FOV remains the same
• Spatial resolution improves as matrix size increases for the same FOV because a larger matrix provides smaller pixels
• Spatial resolution/image sharpness is a key image quality factor determined by pixel size measured in line pairs per millimeter (lp/mm)
• Each pixel matrix can represent a wide range of shades of gray
• Grayscale bit depth determines the number of shades of gray
• Grayscale ranges from 8 bits to 32 bits
• Grayscale bit depth of 8-32 = 1–4 bytes of storage per pixel
• A grayscale bit depth of 12 produces 212 gray levels/4,096 shades of gray
• Density resolution is 4,096 different shades of gray available,
• Digital images are represented in two domains that depend on how the image is acquired:
  • Spatial locations
  • Spatial frequency
• Images in the spatial location domain have specific matrix locations for each pixel.
• Images in the spatial frequency domain are built upon the number of cycles per unit length.
• Spatial frequency is usually described as the object size and contrast
• The use of the Fourier transformation (FT) is a mathematical algorithm applied to change an image from the spatial location domain to the spatial frequency domain
• Physicians and radiographers view images in the spatial location domain. Whereas physicists and engineers can extract information from the mathematical representations of the image in the spatial frequency domain

Image Processing Operations

• Improving diagnostic quality by changing pixel input values is achieved through computer operations
• Image processing operations include
  • Point processing,
  • Local processing
  • Geometric processing
• Point processing operations are the most important for digital radiography imaging systems
• Point processing operations involve grayscale processing for image brightness and contrast adjustments
• Grayscale processing involves:
  • Histogram creation
  • Application of LUT (look-up table)
  • Windowing
• A histogram is generated during initial processing from the image data
• The digital system locates the minimum and maximum signal within the anatomical regions of interest in the image to find the useful signal
• Each scanned area is divided into pixels, and the signal intensity for each pixel is determined to generate a histogram
• The shape of a histogram will correspond to the anatomy and technique
• A histogram analysis is performed by the system to determine the values of interest (VOI) and the exposure indicator
• VOI locates the minimum and maximum exposure values for the body part
• A reference histogram stored in the computer is used to compare histograms to VOI
• Histogram shifting or rescaling to the right area can correct the image if the exposure is outside the range of underexposure or overexposure
• Histogram modification/stretching is changing a histogram by stretching
• A wide histogram demonstrates higher contrast, whereas a narrow histogram shows lower contrast
• Dark/dim images is due to the histogram values being concentrated in lower comparative values
• The next step is to adjust the contrast
• Digital detectors have a linear response that would result in a low-contrast image if no adjustments are made
• An LUT is applied to provide an exam with a standard contrast level for the image to be displayed
• The proper LUT will provide the proper grayscale, regardless of variations in kVp and mAs
• This results in consistent images
• Diagnostic images will not be produced and systems cannot compensate if exposure values are far outside the normal range
• This is why radiographers must set exposure factors that will produce images within normal exposure ranges
• Windowing is a point processing operation that changes the contrast and brightness displayed
• The brightness and contrast of the digital image depends on each pixel's numerical values(shades of gray)
• Images are viewed as a "window" because the range of stored shades of gray is wider than the human eye can see
• Window width (WW) is the range of shades of gray that is displayed
• A narrow width = image will have high contrast
• A larger width = image will have low contrast
• Window level is the center of the window width and controls the brightness of the image
• Levels set at the low end of the scale produce lighter images
• Levels set at the high end produce darker images
• Quantitiy of data stored for each pixel relies on the bit depth imaging abilities of the system
• It is important for radiographers to determine which portion of the imaged information will be displayed
• Adjustments before PACS affect the radiologists ability to diagnose
• Careful image processing is critical to avoid obscuring diagnostically critical information
• Image subtraction and temporal averaging are considered to be point processing operations
• Primarily point processing operations are used in digital subtraction angiography

Local Processing Operations

• Mathematical calculations are applied to a small group of pixels during local processing, even if the the operation extends to all pixels
• A kernel is processing code that is mandatory and common to the computer system, and is repeatedly applied to the set of data being processed
• Spatial frequency filtering is used to:
  • Sharpen
  • Smooth
  • Blur
  • Reduce noise
  • Pull elements of interest
• It can occur in the spatial location domain or the spatial frequency domain
• Fourier transform is used for spatial frequency filtering conducted in the spatial frequency domain
• Pixel values are used for spatial frequency filtering done in the spatial location domain
• High-pass filtering/edge enhancement/sharpness converts the image into the spatial frequency domain
• A high-pass filter is applied to remove low-spatial frequency and produce a sharper output image, structures of interest frequency is amplified
• Low-pass filtering/smoothing blurs the image, reducing noise and the displayed brightness level of the pixels
• Images that are low-pass filtered should also be viewed in a normal or high-pass mode
• Unsharp masking/blurring subtracts a low-pass filtered blurred image from the original image, producing a new subtracted and sharper image
• Spatial location filtering is also known as convolution applying a kernel to weigh pixel vales or apply a coefficient

Geometric processing operations

• Geometric processing changes the position or orientation of the pixels in the digital image
• Allows digital photo operations like rotation and magnification

Digital Image Quality

• Described in terms of spatial resolution, noise, detective quantum efficiency, and artifacts
• Spatial resolution depends on matrix size and how many pixels can be displayed by the monitor
• Matrix and spatial resolution have a direct relationship - Spatial resolution increases If matrix size is increased and pixel size goes down
• Matrix size is controlled by the design of image detectors and the accompanying electronics built into the system
• Density resolution shows the effect of bit depth - Better density resolution is provided when there is greater bit depth
• Contrast resolution is the ability of a detector to resolve different energy differences striking it and transfer these energies into signal values, which are expressed as varying grayscale levels
• Low-contrast resolution is visualizes subtle energy differences, particularly in soft tissues and is often a visual indication of underlying pathology
• CR and DR have superior low-contrast resolution compared to film-screen imaging because of the greater dynamic range resulting in more gray shades and the improved visualization of pathology on digital images
• Noise is a result of:
  • System noise
  • Ambient noise
  • Quantum mottle noise
• System noise is from the digital system itself, often through normal functioning/ background information that does not contribute to image quality
• Measured as the signal-to-noise ratio (SNR)
• An image with little noise is indicated by a high SNR
• Noise decreases when image contrast is increased
• Image noise effects on brightness is irrelevant
• Quantum mottle noise comes from an insufficient quantity of photons from improperly set exposure factors
• Correct by increasing mAs or kVp to produce grainy images which also increases patient exposure
• Using correct exposure produces better-quality images and reasonable patient exposure
• A measure of the sensitivity and accuracy by which the image receptor converts the incoming data to the output viewing device is known as the detective quantum efficiency (DQE)
• A perfect DQE is 1 (100 percent efficiency/no loss of information)
• Digital radiography imaging systems have DQEs of 0.3–0.7 or 30–70 percent.
• DQE is affected by the
  • SNR
  • Quantum noise
  • System noise
• DQE can be described as the ratio of the squared output signal-to-noise ratio (SNRo)2 to the squared input signal-to-noise ratio (SNRI)2 of the imaging detector.
• In addition to SNR and DQE is dependent on modulation transfer function (MTF)
  • Incident x-ray energy
  • Detector material
  • Spatial frequency
• A modulation transfer function measures the capacity of the detector to pass its spatial resolution characteristics to the final image
• DQE and MTF of a digital detector are directly proportional, while DQE and SNR have an inverse relationship

Receptor Exposure

• Less radiation is necessary to achieve identical image quality that a lower DQE detector would produce when the DQE values are high
• A digital receptor with a higher DQE doesn't always provide better image quality at a lower radiation dose
• This is because of other factors like:
  • Incident beam energy
  • Signal sampling and processing
  • Quality of image display
• Exposure indicators reflect the amount of x-ray exposure to an image receptor/quantity of photons striking the detector
• Exposure indicator numbers can be used as a quality control tool for image quality and radiation protection
• Target exposure indicators (EI) provide optimum exposure and acceptable noise levels.
• The correct exposure factors will deliver an acceptable exposure indicator for the best image quality
• The exposure sensitivity of imaging detectors ranges from 0.1 mR up to 100 mR
• Manufacturers calibrate the response of their systems differently and report the results using parameters that have different dependencies on exposure
• Digital radiography processing systems use exposure indicators to provide information regarding the exposure to the image receptor useful in determining patient exposure and ALARA compliance
• There is no universal system
• Fuji, Konica, and Philips computed radiography systems use a sensitivity number (S number) to assist in evaluating exposure
• The exposure reaching the imaging plate is inversely proportional to the sensitivity number (S)
• S=200 (exposure in mR) formula Fuji's sensitivity numbers
• Proper exposures on imaging plates = S values of 150–250
• The Carestream (formerly Kodak) CR indicator system is called the exposure index (EI), which is directly proportional to the radiation striking the imaging plate
• EI depends varies between image receptors which influences beam qualities
• Agfa uses log median exposure log system comparisons
• The deviation index (DI) indicates whether the detector response of a specified image KIND agrees with KTGT