The basic concepts of digital x-ray detectors are covered including the important concepts. Digital detectors are separated into small individual components termed detector elements (DELs), and the size of individual DELs is referred to as the pixel pitch. Whereas the matrix size is the number of DELs in each direction on the detector. The signal range over which the detector can faithfully represent the measured x-rays is the dynamic range. The bit depth is the number of separate computer bits used in saving the value for each DEL.
Digital X-Ray Imaging Sampling Terminology
Here we discuss the terminology relating to the size of each detector element so that when definitions like detector pitch or fill fraction get thrown around you will have a good understanding of the meaning.
Like digital photography, X-ray images are formed with digital elements (DELs). When the image is being stored after it is acquired, or when it is being displayed on a monitor the individual elements are referred to as picture elements (pixels).
Just to keep things clear we use a different terminology to describe the physical detector elements (DELs).
The detector pitch is then defined as the distance from end to end within a DEL. Therefore, smaller DEL size will yield a smaller pitch.
Additionally, within each DEL there is a region that can detect x-rays and an inactive region (such as the electronics of each DEL). The region where x-rays can interact is referred to as the active area. The region that cannot detect x-rays is referred to as the non-active area.
The ratio of active area of DEL to the whole size of each DEL is called “fill fraction”.
The detector fill fraction will therefore be a number that is between 0 and 1. The greater the fill fraction the more x-rays will be captured in the measurements. Thus, a higher fill fraction will be more dose efficient. In general, as the size of each DEL gets smaller the challenge is to ensure that the Fill Fraction stays high since there is associated electronics for each DEL.
Rad Take-Home Point: A digital x-ray detector can be separated into detector elements (DELs), and each element has a fill fraction which indicates the geometric efficiency of the detector to collect the x-ray signal.
The detector matrix is composed of many individual DELs. The matrix size is a two-dimensional number. If matrix size is 1024 x 1024, this means that the matrix has more than one million DELs.
There are also medical flat panel detectors that have matrices of 4288×4288.
We can put these numbers into context we can compare to digital cameras where the matrix size is typically specified in megapixels. A 1024×1024 detector is equal to 1 MegaPixel. A MegaPixel is defined as 220 which is slightly greater than 1 million. The detector that is 4288×4288 is equivalent a 17.5 MegaPixel sensor in terms of the number of Dels.
Rad Take-home Point: The matrix size of a digital x-ray detector indicates how many elements in the whole detector.
Another important characteristic of a digital flat panel detector is the sampling frequency in the detector. This is another way of expressing the size of each Del.
The sampling frequency is inversely proportional to the pixel pitch.
If DELs are larger (i.e. have greater pitch) then the sampling, sampling frequency will be smaller. If DELs are smaller then the sampling frequency will be higher.
Rad Take-home Point: The sampling frequency is inversely related to the Pitch of each Del.
Let’s look at an example to get a feeling of how these parameters are calculated.
If we imagine that our detector’s size is 50cm x 50cm and the matrix size is 1000 x 1000. So, pixel size, according to formula, will be:
So, the sampling frequency will be calculated as follows:
We can calculate the fill factor as well for a sample case. If the DEL size is 1mm x 1mm and size of active area is 0.5mm x 0.5mm, what will the active area be?
This would be a considered a low fill fraction detector as only one quarter of the area of the detector is active and detecting x-rays.
Rad Take-home Point: Just like digital photography, there are a few basic parameters that characterize a digital x-ray detector at a high level.
Digital Sampling Concepts
The majority of x-ray systems in the United States use digital x-ray detectors (indirect or direct conversion flat panel detectors), or so called computed radiography which uses a digital readout as well.
Since you are a Radiologic Technologists or a student and you use or will be soon using these systems many times a day, you are likely interested in the important concepts around how the digital image is formed.
The physics behind the different types of detectors will be covered in a different post but here we focus on the common features amongst all digital x-ray systems.
Rad Take-home Point: The x-rays interact with the detector creating an analog signal, this signal is then converted to a digital signal (a number for each detector element) in the detector.
We will start with an example of the most common x-ray detector in clinical use. In an indirect x-ray detector when X-rays hit the detector they are converted to visible light photons. These light photons are measured by a photo diode which convert them to electrons.
In this case the number of electrons is the analog signal and it is digitized as the electrons go through circuits, and a single number is assigned to each detector element.
The digital circuit converts the energy deposited in the detector to a sampled (i.e. digitized) number. In the figure you can see the effect of the number of bins on the digitization. On the left is the true energy. On the right are the digitized versions of the signal with different bit depths.
When the signal is digitized every bit will be set to either 0 or 1. It is set to 1 if the true signal is above the level and set to 0 if the true signal is below the level.
The number of levels in the digitization is directly related to how many bits the detector has in the analog to digital conversion circuit: number of Levels = 2N , where N is the bit depth.
So, if we use a 4 bit conversion of the energy to digital signal, the accuracy will be much less than in case of 8 or 16 bit conversion. In general, the more levels you have (i.e. the higher the bit depth) the more accurate the image will be.
Rad Take-Home Point: In all digital x-ray detectors an analog signal is converted into a digital signal and the conversion is more accurate when higher bit depths are used.
Dynamic range is also part of the digitization process and is related to the bit depth. The dynamic range is the range over which the signal will be properly digitized. For instance for signals that are higher than then upper end of the dynamic range the signal that is read out will be saturated as it can not handle the high signal levels.
As discussed in the section above there is a desire to make the size of each digitization bin smaller. This can be accomplished by adding more bins as discussed above. The length of each bin can also be reduced by reducing the supported range. This range of supported signal levels is referred to as the dynamic range of the system. The height of each digitization bin is simply: Digital bin height = Dynamic Range / (Number of Bins -1).
In this figure you can see the issues that can occur and why the dynamic range and the bit depth must be chosen carefully. In the figure the dynamic range is changed while leaving the bit depth constant.
If the dynamic range of the system is too small then signals with a very high signal level will be saturated and the true value will not be recorded, rather just the highest value that the system can record with be used.
On the other hand, if the dynamic range is too large then there will be wasted bits in the conversion that are never used, and each bit will cover a greater signal range. Since there is a desire to have each bit cover a smaller signal range, an overly large dynamic range is also not optimal.
In the optimal case the dynamic range of the system will cover most all signal levels that are expected on the system so that saturation does not occur, but it will not be so large that there are significant digitization errors. When the dynamic range is chosen appropriately this is the ‘well sampled’ region in the figure.
In clinical images if the detector does not have a large enough dynamic range the values in areas of very high signal, such as the lungs, will be saturated and structural differences in the lung tissue will be lost.
Rad Take-home Point: The range of all values that are properly digitized is known as the dynamic range of the detector and the dynamic range must be chosen appropriately to reduce the size of each sampling bin, but without having saturation.