British Journal of Radiology (2004) 77, 950-952
© 2004 British Institute of Radiology
doi: 10.1259/bjr/21117919
Technical note: a comparison of antiscatter grids for digital radiography
L Court, PhD
and
T Yamazaki, BS
DR System Development Division, Canon Inc., Utsunomiya, Japan
Correspondence: Tatsuya Yamazaki, DR Systems Development Division, Canon Inc., 20-2 Kiyohara Kogyo-Danchi, Utsunomiya, Tochigi, Japan, Zip 321-3292
 |
Abstract
|
|---|
The use of digital radiography (DR) systems offers a number of advantages over film–screen detectors. One potential disadvantage, however, is that some fixed DR systems do not allow the user to change the antiscatter grid to suit the imaging task. Instead, the user must choose the grid at the time of purchase. Six grids, which are offered as installation options for one commercial fixed-room DR system, were experimentally evaluated, using a range of scatter conditions and tube voltages. In addition, three grids, which are available with a portable DR system in which the user can change the grid to suit the imaging task, were also evaluated. The grids were compared using the primary transmission, scatter fraction, and calculated signal-to-noise improvement factor (SIF). It was found that the grids with low atomic number interspace and cover material had an SIF up to 15% higher than did the grids with aluminium interspace and cover material; the grid with a grid ratio of 12:1 had the highest SIF for all tube voltages and scatter conditions tested here. This 12:1 grid probably represents a good general-purpose non-removable grid in DR.
|
Introduction
|
|---|
The design and selection of antiscatter grids for different imaging tasks has been studied extensively, and the advantages of low atomic number cover and interspace materials are well understood [1–6]. Although ideally the user selects a grid on the basis of the imaging conditions (e.g. examination type or patient size), several of the fixed-room digital radiography (DR) systems currently available have a pre-installed grid that is not easily accessible to the user. This means that for some systems, the user must select a grid at the time of purchase. For example, for one of the DR systems currently on the market (CXDI-11; Canon Inc., Tokyo, Japan), the user must choose one of three grid ratios (8:1, 10:1 and 12:1), and one of two grid cover/interspace materials (aluminium [Al] or low atomic number material/carbon fibre [CF]) with strip densities of 40 cm–1 (Al) and 44 cm–1 (CF), giving the user a choice of six grids. The chosen grid is installed into the system. For this system, the grid lines are removed from the images by moving the grid during the X-ray exposure.
Instead of moving the grid to remove grid lines from the image, several DR systems take the alternative approach of using grids with very high densities of 78 cm–1 (Revolution XR/d; GE Medical Systems, Waukesha, WI) to 80 cm–1 (Thorax FD; Siemens Medical Solutions, Erlangen, Germany). These systems do not offer alternative grid materials, but the differences between aluminium and low atomic number materials are known to be relatively small for high strip density grids.
A third approach to removing grid lines from the digital images is to remove them using software after image acquisition. One DR system that uses this approach is a portable system (CXDI-31; Canon Inc.) which, unlike the fixed DR systems described above, allows the user to change the grid to suit the imaging task. When the software approach to grid line removal is taken, the strip density is chosen on the basis of a complex relationship between the spatial sampling frequency (pixel size) and spatial resolution (e.g. modulation transfer function) of the sensor, and the ability of software to remove grids of different frequencies. For this system (100 µm pixels), this means that a medium-high strip density of 60 cm–1 should be used. The user is supplied with a range of grid ratios, including 4:1, 8:1 and 10:1.
The purpose of this study was to evaluate the six grids for the fixed DR system (CXDI-11), and the three grids for the portable DR system (CXDI-31) described above. It is hoped that our results will assist users in choosing a grid for DR.
|
Materials and methods
|
|---|
The parameters of the grids examined for this study are given in Table 1
. The primary transmission (ratio of detected signal due to primary X-rays when a grid is used to that when no grid is used) and scatter fraction (fraction of the detected signal which is due to scattered X-rays) of each grid was measured using a CXDI-11 DR system, using experimental techniques described by Chan et al [5, 6]. Specifically, the primary transmission was evaluated using narrow-beam geometry; scatter fractions were measured using a lead disk method (6 lead disks, 2 mm thick, with diameters ranging from 20 mm down to 2 mm). Measurements were made using X-ray tube voltages of 80 kVp, 100 kVp, and 120 kVp, using Perspex thicknesses of 5 cm, 10 cm, and 15 cm to simulate different scattering conditions. The source-to-detector distance was set equal to the grid focal distance (110 cm, 140 cm, or 180 cm, depending on the grid). The Perspex phantoms were positioned immediately adjacent to the detector cover, thus giving a geometry representative of that used in patient imaging. When measuring scatter fraction, the field size at the sensor was set to 30 cm x 30 cm. For a given Perspex thickness and tube voltage, the exposure was kept constant.
View this table:
[in this window]
[in a new window]
|
Table 1. Parameters and experimental results of the nine grids. DR system indicates the type of DR (digital radiography) system for which these grids are available
|
|
Because image contrast can be digitally altered in digital radiography, the signal-to-noise ratio (SNR) of an image due to a low-contrast object is the appropriate metric for comparisons of image quality. The experimentally measured primary transmission and scatter fraction for each grid can be used to estimate the relative SNR of a low contrast object, using the following approximation:
where S is the scatter fraction, and Tp is the primary transmission. The derivation of this formula is similar to that in Chan et al [6]. The constant of proportionality is dependent on the signal per scattered or primary photon incident on the detector. If it is assumed that the use of different grids does not affect this significantly, then different grids can then be compared using the SNR improvement factor [5], SIF, which is the ratio of the calculated SNR with a test grid to the SNR with a reference grid. SIF, therefore, represents the improvement in image quality that can be achieved if the reference grid is replaced by the test grid (with no change in exposure). Here, grid 2 (Al interspace material, 10:1 grid ratio) was chosen as the reference grid, as it is one of the most common grid types in general radiography, and was the grid originally installed in the DR system used for this evaluation.
|
Results and discussion
|
|---|
The primary transmission and scatter fraction found with the nine grids, using 5 cm and 15 cm Perspex as the scattering material, are given in Figure 1. This represents the range of scatter conditions tested here. The experimental errors in primary transmission and scatter fraction were estimated at better than 1–2%. The contrast improves as the scatter fraction decreases, so grids towards the left of Figure 1 will give images with higher inherent contrast (i.e. contrast before software adjustment) compared with grids towards the right of Figure 1. The primary transmission is proportional to the square of the SNR that will be achieved for zero scatter conditions, so the effect on patient dose or image quality of not being able to remove the grid for examinations that do not normally require a grid will be less for grids towards the top of Figure 1 compared with grids towards the bottom of Figure 1. The SIF calculated from these experimental results are given in Table 1 for tube voltages of 80 kVp and 120 kVp, again representing the range of conditions tested here.
First, we compared the grids for the fixed DR system (grids 1–6). For all scatter conditions the 12:1 CF grid (grid 6) gave approximately the same scatter fraction as the 10:1 Al grid (grid 2), whereas the 10:1 CF grid (grid 5) had slightly worse scatter fraction than the 10:1 Al grid. This means that the 12:1 CF grid would give approximately the same contrast as the 10:1 Al grid, and the 10:1 CF grid would give a slightly worst contrast that the 10:1 Al grid. For the lowest scattering conditions evaluated here (5 cm Perspex), the largest SIF values were found with the 10:1 and 12:1 CF grids. For medium and high scattering conditions (10 cm and 15 cm Perspex), the 12:1 CF grid gave the best SIF, although the difference between the 10:1 and 12:1 grids was only 1–4%, and would probably not be clinically noticeable. These small differences in SNR estimated using Equation 1 may also be susceptible to approximations inherent in this equation, including the effect of spectral differences on the SNR. SIF for zero scatter conditions can be estimated as the square root of the primary transmission factor. This calculation shows that, when compared with no grid, the CF grids will reduce the SNR by 13–16%, and the Al grids will reduce the SNR by 20–28%.
We then compared the results for the three medium-high density grids (grids 7–9) used with the portable DR unit. As expected, these grids had poorer SIF than the standard 10:1 Al grid. This is a result of their relatively high strip density, which leads to poorer primary transmission or poorer scatter rejection, depending on the grid ratio. These grids had SIFs 2–17% lower than the reference grid. From these three grids, the 10:1 grid gave the lowest scatter fractions (and therefore, the highest contrast). The 8:1 grid, however, consistently gave the highest SIF, and thus is probably a good general purpose grid. However, because the DR system that uses these grids is a portable unit and is not physically (or electromechanically) attached to the X-ray tube, alignment of the grid may be difficult, in which case the 4:1 grid may be useful. Note that some authors [4] have assumed that an image SNR up to 10% lower than the SNR of a reference image will still fulfil its diagnostic purpose. The 4:1 grid gave an SIF within 11% of that of the 8:1 grid for all conditions investigated here, meaning that, for a digital system, the 4:1 grid will probably give diagnostic quality close to that achievable with the 8:1 grid.
|
Conclusions
|
|---|
Nine grids that are available for current DR systems were experimentally characterized, and the expected impact on image quality calculated as a SNR improvement factor, SIF. From the three grids available for a portable DR system, the 8:1 grid consistently gave the highest SIF, although the 4:1 grid gave SIF values within 11% of those of the 8:1 grid. From the six grids available for a fixed DR system, the grid with low atomic number interspace material, and a 12:1 grid ratio, gave the highest values of SIF for all scatter conditions, and thus appears to be the best option for a general-purpose non-removable grid in DR.
|
Footnotes
|
|---|
Current address for Dr Laurence Court, Department of Radiation Physics, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd. Unit 94, Houston, TX, USA. 
Received for publication September 22, 2003.
Revision received June 23, 2004.
Accepted for publication July 8, 2004.
|
References
|
|---|
- Sandborg M, Dance DR, Alm Carlsson G, Persliden J. Selection of anti-scatter grids for different imaging tasks: the advantage of low atomic number cover and interspace materials. Br J Radiol 1993;66:1151–63.[Abstract/Free Full Text]
- Sandborg M, Dance DR, Alm Carlsson G, Persliden J. Monte Carlo study of grid performance in diagnostic radiology: factors which affect the selection of tube potential and grid ratio. Br J Radiol 1993;66:1164–76.[Abstract/Free Full Text]
- Sandborg M, Dance DR, Alm Carlsson G, Persliden J. Monte Carlo study of grid performance in diagnostic radiology: task dependent optimization for screen–film imaging. Br J Radiol 1994;67:76–85.[Abstract/Free Full Text]
- McVey G, Sandborg M, Dance DR, Alm Carlsson G. A study and optimization of lumbar spine X-ray imaging system. Br J Radiol 2003;76:177–88.[Abstract/Free Full Text]
- Chan H-P, Doi K. The validity of Monte Carlo simulation in studies of scattered radiation in diagnostic radiology. Phys Med Biol 1983;28:109–29.[CrossRef][Medline]
- Chan H-P, Lam KL, Wu Y. Studies of performance of antiscatter grids in digital radiography: effect on signal-to-noise ratio. Med Phys 1990;17:655–64.[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
B. J. Irving, G. J. Maree, E. R. Hering, and T. S. Douglas
Radiation dose from a linear slit scanning X-ray machine with full-body imaging capabilities
Radiat Prot Dosimetry,
July 1, 2008;
130(4):
482 - 489.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
: 12:1, 
: 10:1, x : 8:1, 
: 4:1.