Originally, screen-film radiography (SFR) was used in which a physical copy of the x-ray film was produced. These have now been replaced by digital radiography. There are two different techniques: computed radiography and digital radiography.
1. Computed radiography
Cassettes are used that have a phosphor screen. When the x-rays hit they form a latent image in the phosphor. The cassette is then placed into a reader with a laser shone on to it which releases the stored photons, collects the signal, and digitises it to be displayed on a display screen.
2. Digital radiography
With digital radiography no cassettes are used. The x-rays hit a permanently placed set of hardware, which then sends the digital information directly to a readout mechanism.
- Indirect DR: x-ray photons hit a scintillator layer, which then releases light photons that then hit an active matrix array that digitises the signal
- Direct DR: x-ray photons act directly on a photoconductor layer producing positive and negative charge. The positive charge is attracted to a charge capacitor that stores the latent image. It is then read out by TFT switches pixel by pixel.
Standard DR process
- X-ray produced by standard radiographic x-ray tube
- Image captured by digital image detector
- Digitised into a stream of data via an analogue-to-digital converter (ADC)
- Transfer to a system computer
- Output via digital-to-analogue converter (DAC) to video format
- Post-processing of image
- Display on to suitable display device
Computed radiography (CR)
X-ray luminescence
X-ray luminescence is the physical mechanism by which x-ray energy is converted into light in a phosphor screen. It involves two mechanisms that both occur to some degree when a phosphor screen is irradiated:
- X-ray fluorescence: the immediate emission of light. This is the mechanism that predominates in screen film radiography
- X-ray phosphorescence: this is when the emission of light is delayed over a timescale of many minutes, hours or days and can be accelerated by shining specific coloured light onto the phosphor. This is the mechanism exploited in CR. It allows x-ray energy to be temporarily stored in a phosphor screen to be read-out later.
CR image plate (CR IP)
The plate is a layer of phosphor crystals (made of barium fluorohalide activated with divalent europium ions (BaFX:Eu)) embedded in a polymer binder with the top surface protected by a layer of toughened plastic. It is typically 0.3 mm thick.
| Standard IP | High resolution IP | |
| Layer of phosphor crystal | Thicker layer | Thinner layer |
| Crystal size | Larger | Smaller |
| Light reflection layer | Yes | No |
| Uses | General radiographic examinations | High spatial resolution |
| Fractional x-ray absorption efficiency | 40% (good) | Lower i.e. need larger x-ray dose |
Image processing
1. Latent image formation
X-ray photons are absorbed into a phosphor crystal giving rise to a high energy photoelectron. This ionises a large number of atoms along its track releasing thousands of electrons (one x-ray photon absorbed gives rise to over 100 trapped electrons). The electrons become temporarily trapped at specific sites throughout the layer of phosphor crystals producing the latent image.
2. Laser simulated emission
If left long enough the electrons spontaneously relax back to their ground state and the image decays over time. During readout the IP is scanned with a red laser beam stimulating the trapped electrons to immediately relax back to their ground state and release their stored energy as light photons in the blue part of the spectrum. The light photons are then collected by optical fibres to a photomultiplier (PM) tube. The PM tube produces an electrical current.
3. Resetting cassette
Readout is “destructive” as it eliminates the latent image. The film is then exposed to bright light to erase any residual signal before re-using the cassette.
4. Post-processing of image
Digital image structure
Pixel
Spatial resolution is determined by pixel size. Each pixel records a value, in binary format, related to intensity of signal in the corresponding part of the image. In binary system 1 bit is one value of grey.
N bits = 2n (number of different values of grey)
Computer memory is measured in bytes:
1 byte = 8 bits (28 = 256 values)
Image quality
Exposure Index (speed)
The Exposure Index (EI) is a measure of the amount of exposure on the image receptor. In screen-film radiography it is clear if the image is under- or overexposed as it will be too bright or too dark. In computed or digital radiography the image brightness is altered digitally and there is no longer a clear visual link. However, if an image is under or overexposed this can still affect the image quality by introducing noise or reducing contrast. Manufacturers measure how ideal the exposure is with the EI. Each manufacturer provides a recommended EI range for optical quality. Each manufacturer has a slightly different method of measuring EI.
An example, one manufacturer assesses EI by calculating a “sensitivity number (S-number)” which is calculated as:
The S-number usually operates from 200-300.
- S < 200 improved signal to noise ratio but increased patient dose
- S > 400 used when minimal radiation required e.g. repeated paediatric films
Latitude (dynamic range)
Unlike SFR (which has a characteristic curve), the dynamic range is very high and the dose-response is linear meaning CR produces good contrast over a much wider range of exposures.
Spatial resolution
Improved by:
- Smaller diameter of readout laser beam (thinner line of image plate “read out”)
- Smaller pixels
- Smaller size of phosphor crystals
- Thinner phosphor layer
- No light reflection / absorption backing layer (as this produces scatter despite improving efficiency by using more of the photons for image production)
Spatial resolution is best described by the modulation transfer function (MTF).
Modulation transfer function
The MTF represents the ratio of output to input modulation. An MTF of 1 means the spatial resolution imaged and displayed are the same. As the spatial frequency increases the MTF decreases until, with the addition of noise, it is impossible to visualise details of higher spatial frequencies – the “limiting spatial resolution” – and the MTF is 0 (i.e. no information conveyed).
Detective quantum efficiency (DQE) of CR imaging
This is defined by the follow equation:
The higher the DQE the more efficiently the detector can record information. A DQE of 0.25 implies that the detector can only exploit ¼ of the incident x-ray photons. For a CR imaging system it is typically:
- 0.25 for a standard IP
- 0.12 for high resolution IP
Artefacts
Moiré pattern: when a stationary x-ray anti-scatter grid is used there is interference between the linear structure of the grid and the regular pixel array of the digitised image.
Ghost image: due to carry-over of image content from a previous exposure.
Excessively high / low image density: due to faulty operation of the data auto-ranging software, previously due to incorrect identification of the x-ray collimators.
Excessive digital enhancement: e.g. ringing effects along the edges of high density structures or shadowing within such structures.
Written by radiologists, for radiologists with plenty of easy-to-follow diagrams to explain complicated concepts. An excellent resource for radiology physics revision.
Digital radiography
In CR the film cassette has to be removed from under the patient and fed into a reader to be processed. In digital radiography (DR) the image is produced directly from the image detector and is displayed on the screen.
There are two types:
- Indirect DR: x-ray → stored electrons → light photons → readout electronics
- Direct DR: x-ray → charge → readout electronics
Indirect DR
Hardware
1) Scintillator layer
Most systems use a thin 500 μm layer of caesium iodide (CsI:TI) as a scintillator to capture the image which is coated onto the hydrogenated amorphous silicon (a-Si:H) active matrix array (some systems use gadolinium oxysulfide as the scintillator layer). The CsI:TI is a channeled crystal structure that ensures minimum unsharpness caused by scatter of the recorded image. Absorption of an x-ray photon releases ~3000 light photons in the green part of the spectrum.
2) Active matrix
This is formed by a layer of a-Si:H and forms the readout electronics. The active matrix consists of a high resolution array of electronic components. Each pixel typically comprises a:
- Photodiode (a light sensor) – amplifies signal from incident light photons
- Charge storage capacitor – stores signal of latent image
- Thin-film transistor (or TFT switch) – latent image read out and transferred to TFT switches that produce a voltage signal that is digitised and converted into the image
This circuitry (TFT and charge storage capacitor) takes up a small area of each pixel preventing image formation in this area. This is calculated by the fill factor.
Fill factor = sensitive area / overall area
Decreasing the pixel size (making each area smaller) improves the resolution but, as the circuitry remains the same size, the fill factor and, therefore, the efficiency of the array, decreases.
3) TFT array
This is a device that amplifies the signal then stores it as an electrical charge. The charge can be released and read by applying a high potential. In the array each transistor corresponds to a pixel.
4) X-ray window
The translucent x-ray window is made of aluminium or carbon fibre over the detector entrance to minimise unnecessary absorption and scatter of x-ray photons.
Image formation
- CsI:TI absorbs x-ray photons and releases light photons
- These light photons are then absorbed in the photodiodes and the charge stored in the charge storage capacitor at each pixel location
- The latent image is read out sequentially to a bank of charge sensitive amplifier (TFT switches)
- The resulting voltage signal is then digitised and transferred to the system computer where the DR image is built up
Direct DR
A layer of x-ray photoconductor material is used instead of an x-ray scintillator.
Photoconductor
This directly converts x-ray photon energy into free electrical charge carriers (electrons and holes) i.e. the “middle-men” or light photons, are cut out. The most commonly used photoconductor is amorphous selenium (a-Se).
Sequence of image formation
- X-ray photon absorbed by a-Se photoconductor
- Electrical charge carriers (negative electrons and positive holes) are created in the a-Se
- A surface electrode at positive potential attracts and discards all the electrons
- The positive charges are drawn to the charge storage capacitor forming the latent image
- The latent image is then read out sequentially by gating each row of TFT switches (each TFT corresponds to one pixel) in turn to read the charge pattern and transfer to a bank of charge sensitive amplifiers
- The resulting voltage signal is then digitised and transferred to the system computer where the DR image is built up
- Post-processing
Post-processing
Artefacts and correction
Artefacts
- Irregular shading across field: due to non-uniform variations in the sensitivity or gain of the x-ray absorption layer
- Bright / dark spots or lines in image: due to individual rows and/or columns of defective pixels in the active matrix array
Correction
- Gain calibration: uses previously acquired mask image comprising an image acquired with a uniform x-ray beam and subtracting this gain mask image from the patient’s image
- Pixel-calibration: defects in pixel array can be corrected by interpolating the data values of neighbouring pixels which are functioning correctly using a reference map
Auto-ranging
The data needs to be matched to the display device.
- Identification of relevant image field
- Generation of a histogram of the data representing the number of pixels at each grey-scale value
- Analysis of the histogram to exclude ranges of data which contain no clinical information (very high and low values)
- Selected grey-scale range normalised to match the display image
Digital image enhancement
Grey-scale modification
A look-up-table (LUT) is a method of systematically re-mapping the grey-scale values in the recorded image to a new range of values in order to improve the displayed image in some way. Shifting the LUT gradient and position adjusts the mean brightness and displayed contrast of the image.
Spatial feature enhancement
- An unsharp mask algorithm is used to produce a blurred version of the original image
- This is then subtracted from the original image to produce an image which retains only the fine detail structures in the image
- Add the fine detail image back onto the original
- Produces enhanced composite image
Monitor display
Cathode ray tube (CRT)
Visible image generated by scanning a phosphor screen with a focused beam of electrons all contained within an evacuated glass tube.
Flat panel displays
Most display monitors are based on liquid crystal technology. Application of the appropriate voltage distribution to an active matrix modulates light polarisation on a pixel-by-pixel basis varying the light emission that comprises the image seen on the screen. It produces a higher contrast image with greater resolution and less power usage.
Hardcopy
On occasions it is necessary to print a hardcopy image. A hardcopy image is recorded using a laser printer onto a film with silver crystals to create a latent image. This is converted into a visible image by applying heat to the film. This ‘dry’ film processing eliminates the need for traditional chemical processing.
Σ Summary
Computed radiography (CR)
- Image formed on phosphor cassette that is removed, read and then reset to be used again
Process
- X-ray photons absorbed by phosphor crystal
- High energy photoelectron released which ionises atoms along its track releasing electrons → >100 electrons released per x-ray photon
- Cassette removed and placed in machine for read-out
- Red laser beam scans back and forth releasing energy from electrons, which is released as blue light
- Light collected by optical fibers to PMT
- PMT produces electrical current
Image quality
- Exposure Index (speed)
- Manufacturers calculate a ‘sensitivity number (S-number)’ to measure this
- Typically S < 200 → improved SNR but at increased patient dose
- Typically S > 400 → for when minimal radiation required
- Latitude
- Dynamic range is a straight line = good contrast over wide range of exposures
- Spatial resolution
- Described by modulation transfer function (MTF): 1 = spatial resolution of image is same as of object. 0 = no information in the image
- Improved by:
- Smaller readout laser beam
- Smaller pixels
- Thinner phosphor layer
- Smaller phosphor crystals
- No light reflection / absorption backing layer
- Detective quantum efficiency (DQE)
- Measure of sensitivity of detector
- DQE = SNR2out / SNR2in
Digital radiography (DR)
Indirect DR: x-ray photons → light photons → electrical signal
- Process:
- X-ray photon hits CsI:TI scintillator layer releasing ~3000 green light photons
- Light photons detected by active matrix of a-Si:H which is separated into pixels with each pixel containing a photodiode and charge storage capacitor
- Photodiode – amplifies signal
- Charge storage capacitor – stores signal of latent image
- TFT switch – latent image read out and transferred to TFT switches that produce voltage signal that is digitised and converted into the image
- Fill factor: TFT and charge storage take up small area of pixel. Fill factor = sensitive area / overall area
Direct DR: x-ray photons → electrical signal
- Process:
- X-ray photon absorbed by a-Se photoconductor
- Electrical charge carriers created. The positive charges are drawn to the cathode charge storage capacitor to create latent image
- Latent image read-out via TFT switches and transferred to bank of charge sensitive amplifiers
- Voltage signal digitised
Post-processing
- Artefacts:
- Irregular shading due to non-uniform variation in sensitivity or gain
- Bright / dark spots due to individual row / column of defective pixels
- Correction of artefacts:
- Gain-calibration uses mask image obtained with uniform x-ray beam to correct patient image
- Pixel-calibration uses values of neighbouring pixels to correct defects in pixel array
- Auto-ranging:
- Analysis of histogram of image grey-scale data to reject very high and low values that contain no clinical information
- Digital image enhancement:
- Grey-scale modification – look-up-table (LUT) to remap grey-scale values and improve displayed image
- Spatial feature enhancement to produce enhanced composite image
