British Journal of Radiology (2003) 76, 301-308
© 2003 British Institute of Radiology
doi: 10.1259/bjr/32889803
X-ray refraction effects: application to the imaging of biological tissues
R A Lewis, PhD
1
C J Hall, PhD
1
A P Hufton, PhD
2
S Evans, MSc
3
R H Menk, PhD
4
F Arfelli, PhD
5
L Rigon, MS
4
G Tromba, PhD
4
D R Dance, PhD
3
I O Ellis, FRCPath
6
A Evans, FRCR
6
E Jacobs, MSc
7
S E Pinder, FRCPath
6 and
K D Rogers, PhD
8
1 Daresbury Laboratory, Warrington WA4 4AD, UK, 2 North Western Medical Physics, Christie Hospital NHS Trust, Manchester M20 4BX, UK, 3 Joint Department of Physics, The Royal Marsden NHS Trust, London SW3 6JJ, UK, 4 Sincrotrone Trieste Società Consortile per Azioni, Basovizza, Italy, 5 Department of Physics, Università di Trieste, Italy, 6 City Hospital, Nottingham NG5 1PB, UK, 7 Cancer Studies Unit, Queens Medical Centre, Nottingham, UK and8 Department of Materials and Medical Sciences, Cranfield University, Swindon SN6 8LA, UK
 |
Abstract
|
|---|
The purpose of this study was to explore the potential of refraction contrast X-ray imaging of biological tissues. Images of dissected mouse lungs, heart, liver and legs were produced using the medical beamline at the Elettra Synchrotron at Trieste, Italy. The technique used was diffraction enhanced imaging. This utilizes a silicon crystal positioned between the tissue sample and the detector to separate refracted X-rays from transmitted and scattered radiation by Bragg diffraction. The contrast in the images produced is related to changes in the X-ray refractive index of the tissues, resulting in remarkable clarity compared with conventional X-ray images based on absorption effects. These changes were greatest at the boundaries between different tissues, giving a marked edge enhancement effect and three-dimensional appearance to the images. The technique provides a way of imaging a property of biological tissues not yet exploited, and further studies are planned to identify specific applications in medical imaging.
|
Introduction
|
|---|
Conventional X-ray imaging relies on differences in linear attenuation coefficients between tissues to produce differences in photon fluence incident upon the detector. Whilst these differences are substantial between bone and soft tissue, for example, they are usually small between the different types of soft tissue. The result is that the inherent contrast between soft tissues is low. Contrast, or signal to noise ratio, can be maximized by using low X-ray energies, but the radiation dose to the patient places practical lower limits on the energies that may be used.
In recent years, techniques have been developed which enable differences in the refractive index of different tissues to be exploited as an imaging tool. Although the refractive index for biological tissues differs from unity by only a few parts per million at X-ray energies, and the difference between different tissues is about 100 times smaller than this [1], the phase shift cross section is of the order of 100 to 1000 times greater than the absorption cross section, resulting in a potentially much more sensitive imaging method [2]. Three techniques can be used for visualizing phase information: interferometry [3]; Fresnel diffraction, or phase-contrast imaging [4]; and diffraction enhanced imaging (DEI) [5, 6]. The latter method was the one used to produce the images presented in this paper. It utilizes Bragg diffraction from a silicon crystal placed between the object and detector to isolate the refracted X-rays, thus allowing pure absorption images and pure refraction images to be produced, rather than a mixture of the two. A synchrotron source was used to provide a quasi-parallel monochromatic X-ray beam of sufficient brightness. Further technical details of the technique have been published by Zhong et al [7].
To date, relatively little work has been carried out using DEI to image biological tissues, and its potential for medical imaging has not yet been fully explored. The technique has been used to image samples of breast tissue [8, 9], and images of whole mice have also been published [5, 7, 10]. Our aim was to investigate the detailed appearance of images produced by the DEI technique in specific organs of the mouse, namely the heart, lungs, liver and legs, as an initial step in identifying areas of clinical radiology that might benefit from the new type of information displayed in these images.
|
Materials and methods
|
|---|
Samples
We chose to image the lungs, liver, heart and leg dissected from a mouse. Our aim was to compare DEI images with conventional X-ray images in order to try and identify what new clinical information could be visualized, and point the way to possible applications in clinical radiology. The limbs were taken from a tumour model representing metastasis of breast cancer to bone developed using EMT6 murine mammary carcinoma in Balb/c mice [11]. Cells from in vitro culture were injected into the left ventricle of the heart in female Balb/c mice aged 4 weeks to 5 weeks. Mice were terminated on loss of condition after 10–14 days. The samples were prepared by removing the skin, resecting the limbs with pelvis attached and trimming away excess muscle. The lungs, liver and heart were from "retired breeders", and all samples were fixed in formal saline. The animal work conformed to United Kingdom Co-ordinating Committee for Cancer Research (UKCCCR) guidelines. For imaging, the samples were held in place in the X-ray beam between two very thin elastomer sheets.
Description of DEI technique and imaging set-up
Images were recorded on the SYRMEP beamline at the Elettra Synchrotron radiation facility in Trieste, Italy [12]. The DEI technique has been described in detail by Chapman et al [6], but a short summary is given here for completeness. Figure 1
illustrates the arrangement for acquiring diffraction enhanced images. A monochromatic beam of X-rays from the synchrotron passes through the sample and falls upon a single silicon crystal known as an "analyser". This crystal will diffract only those X-rays which impinge upon it within a very narrow range of angles. The graph of intensity of diffracted X-rays versus angle of the analyser crystal is known as the "rocking curve", and is approximately Gaussian in shape. The diffracted X-rays are then incident on the detector, producing an image. It is this process that originally gave rise to the term "diffraction enhanced imaging". The analyser reflectivity is close to unity at the Bragg angle 
B and has a rocking curve width at half maximum (Darwin width), 
D, of 
20 µrad for Si(111) at 17 keV, which was the energy used in this investigation. Therefore, although the deviation of the X-rays by refraction is small, typically a few microradians, the narrow rocking curve width of the analyser renders this system sensitive to these tiny changes in angle. If the analyser angle is set half way down the rocking curve (B±D/2), then its reflectivity is 50% for X-rays that transit the object undeviated. Rays that are deviated in the object through small angles will be reflected with either a greater or lower efficiency depending on the angle of incidence relative to the rocking curve. The steep slope of the rocking curve acts as a contrast amplifier and the technique is sensitive enough to reveal X-ray refraction effects that carry different information to conventional absorption contrast images.
The images recorded at the peak of the rocking curve of the analyser crystal represent absorption contrast images that are almost entirely free of scattered radiation, as long as the scatter angle is substantially larger than the width of the rocking curve. This effect is called extinction. However, radiation deflected through angles within the angular range of the rocking curve contributes to the image. The two images recorded with the analyser set to the 50% reflectivity points on either side of the rocking curve, where the variation in reflectivity with angle is of opposite slope, can be combined to produce images known as the apparent absorption image and refraction image, using the following algorithms developed by Chapman et al [6].
The apparent absorption image is calculated using:
The refraction angle image is calculated using:
IL and IH are the intensities in the low and high angle images, respectively, and dR/d are the gradients of the rocking curve at the respective angles. R(L) and R(H) are the reflectivities of the crystal at angles L and H. The algorithm is applied pixel by pixel to the images obtained on the high and low angle sides of the rocking curve. For images obtained at exactly the 50% reflectivity points, where R(L)=R(H) and dR/d (H)=-dR/d (L), the apparent absorption image is simply the sum of the low and high angle images, and the refraction image is the difference.
Both of these new images are somewhat noisier than the images from which they are created, since each is formed from two images. To a first order approximation, the apparent absorption image is scatter free. Contrast is produced by absorption and extinction only. The refraction image contains information which is entirely different from absorption contrast and represents the spatial gradient of refractive index.
The SYRMEP beamline used for this study provides a monochromatic fan beam of X-rays typically 1 mm high and up to 100 mm wide, and tuneable from 8.5 keV to 35 keV. Images were recorded by scanning both the detector and object through the beam to form a two-dimensional image. Images were recorded with a Photonics Science phosphor coupled charge-coupled device (CCD)<1query id="1"> detector (Hystar 2048) having 2048 x 2048 14 µm2 pixels (overall resolution 40 µm) and an active area of 28 mm x 28 mm (Photonics Science Ltd, Robertsbridge, UK). Exposure times were typically 30 s, but could be adjusted by varying the height and intensity of the incoming beam by means of a precision slit system and calibrated absorbers, and by changing the speed of the scanning devices. Ionization chambers were positioned before and after the analyser crystal, in order to measure the rocking curve.
An X-ray energy of 17 keV was used for all the mouse organ images. What we have called contact, or conventional, images of the mouse tissue samples were obtained by positioning the samples close to the detector. However, it should be noted that these images were produced using monochromatic X-rays, and so are already of a higher quality than conventional transmission images produced by a bremsstrahlung spectrum from an X-ray tube.
|
Results
|
|---|
Four types of images are displayed in this paper: contact images; peak of rocking curve images; refraction images and images taken on the positive slope of the rocking curve at 10% reflectivity. The figure of 10% was to some extent arbitrary, but was designed to observe the effect on the image of setting the analyser crystal angle to a larger value, without reducing the transmitted photon flux too much. The entrance air kerma at the position of the samples, as deduced from the ionization chamber readings, was approximately 20 mGy for all images acquired. Since the refraction image is constructed from two images either side of the rocking curve, as previously described, the total dose to the sample is doubled for these images. While these doses may be considered quite high for the samples being investigated, it was not the purpose of this work to try and optimize dose. Further work is planned to address this important area of dose optimization. Also, some additional comments on dose can be found in the following discussion section. Image noise is also dependent on the reflectivity of the analyser crystal, and hence increased noise is evident in the refraction and 10% reflection images. The apparent absorption images are not shown, since their appearance is almost identical to that of the peak images. As can be seen from the figures, the sets of images for the different organs all display similar characteristics. The peak of reflection images are much clearer than the contact images, partly as a result of the improved contrast, due to extinction effects, and partly due to the slight edge enhancement. The refraction images show a marked edge enhancement effect, with one edge of a structure appearing lighter and the other darker than the surrounding tissue, resulting in a three-dimensional appearance. Since DEI is sensitive to changes in refractive index, rather than the absolute value, the refraction images lack large area contrast.
Figure 2 shows images of the mouse liver. The peak of reflection image (Figure 2b) demonstrates branching structures within the liver parenchyma more clearly than the contact image (Figure 2a). These vessels are seen even more clearly on the refraction image (Figure 2c); the peripheral branching being seen extremely clearly down to approximately 50 µm. It seems likely from the configuration of the central vessels that at least some of these structures represent portal veins.
View larger version (101K):
[in this window]
[in a new window]
|
Figure 2. Images of a mouse liver: (a) conventional image; (b) image taken at the peak of the rocking curve; and (c) refraction image. Zoomed sections of the peak and refraction images give a clearer depiction of the level of detail present.
|
|
The lung images in Figure 3 demonstrate the general features already described. Both the peak of reflection and refraction images clearly show lobular, segmental and subsegmental bronchi. The peak image (Figure 3b) also shows areas of aerated lung parenchyma distinct from the fluid-filled areas. Small cystic spaces measuring approximately 50 µm are visible on close inspection. Although the appearances are suggestive of the visualization of individual alveoli, there is currently some debate about the interpretation of such features. It has been suggested that, at least in phase contrast images, this appearance is the result of multiple refractions within several layers of alveoli, and it seems unlikely that individual alveoli can be observed [13]. These features are seen even more clearly and down to a smaller diameter in the refraction image (Figure 3c). The final type of image is the 10% reflectivity image, which was recorded with the analyser set to reflect just 10% of the incident radiation (Figure 3d). It shows a quite different appearance to the peak and refraction images, with the alveoli appearing dark rather than bright. This is probably due to multiple scattering effects in the alveoli, resulting in X-rays deflected through a larger angle. At the 10% reflectivity position, where the analyser is set to a higher angle, these X-rays will be transmitted more readily, resulting in a greater intensity on the image detector.
View larger version (156K):
[in this window]
[in a new window]
|
Figure 3. Images of mouse lungs: (a) conventional image; (b) image taken at the peak of the rocking curve; (c) refraction image; and (d) image taken at the 10% reflectivity point on the rocking curve. Zoomed sections of the peak and refraction images give a clearer depiction of the level of detail present.
|
|
The images of the heart (Figure 4) appear to be less informative than those of the liver and lungs. Nevertheless, the peak image appears sharper and has improved contrast compared with the conventional image. There is evidence of a number of vascular structures, probably representing coronary arteries and cardiac veins, over the surface of the heart in the peak image.
View larger version (95K):
[in this window]
[in a new window]
|
Figure 4. Images of a mouse heart: (a) conventional image; (b) image taken at the peak of the rocking curve; and (c) refraction image.
|
|
The peak of reflection image of the mouse leg (Figure 5b) demonstrates good corto-medullary differentiation but with minimum trabecular detail. A lytic metastasis is visible in the proximal tibia in the metaphysis immediately adjacent to the epiphysis. The refraction image (Figure 5c) shows marked edge enhancement, resulting in the clear depiction of the edge of the bone and enhanced corto-medullary differentiation. Also the trabecular detail compared with the conventional image (Figure 5a) is markedly enhanced. The wide dynamic range of both images, showing bone and soft tissue detail, may be noted.
View larger version (109K):
[in this window]
[in a new window]
|
Figure 5. Images of a mouse leg: (a) conventional image; (b) image taken at the peak of the rocking curve; and (c) refraction image. Zoomed sections of the peak and refraction images give a clearer depiction of the level of detail present.
|
|
|
Discussion
|
|---|
As has been demonstrated, DEI can produce images in which the contrast is effectively due to pure absorption in tissues, or to changes in their refractive index. Although these two extremes have been illustrated, it is possible to construct images containing variable proportions of absorption and refraction. Although the refraction images depict strong edge enhancement, this is not gained at the expense of increased image noise, as would generally be the case when spatial filtering is used to "sharpen" images. It may be noted that, apart from some small adjustments to contrast and brightness to ensure acceptable reproduction, none of the images presented have been computer processed. While both conventional and DEI images may benefit from more aggressive processing, we felt that to do so would obscure the connection between the appearance of the images and the underlying physical processes.
It is difficult from this work on dissected mouse organs to estimate what exposures would be required to give satisfactory images in humans. However, studies using a TOR(MAM) image quality phantom (Leeds Test Objects Ltd, Leeds, UK) at realistic mammographic exposures has shown that DEI can give equivalent image quality to conventional techniques, but at much lower doses [14]. It may also be feasible to lower the dose by increasing the X-ray energy, since the phase component of the refractive index decreases much more slowly than the absorptive component as the X-ray energy increases [2].
A synchrotron radiation source is ideal for investigating DEI and similar new imaging techniques, since it can provide an intense beam of spatially and temporally coherent X-rays. While it is feasible to use such a source for developing new techniques and undertaking a certain amount of clinical research, it would not be suitable for large-scale clinical applications, owing to a variety of factors such as limited availability and geographical considerations. It is probable that in the future such limitations can be overcome through the development of compact, intense monochromatic X-ray sources, which would be suitable for medical imaging [15]. Even now it is possible to observe limited phase contrast effects using conventional X-ray equipment, as has been shown by Kotre and Birch in mammography [16].
From the work presented here it is possible to identify a number of areas of clinical radiology where DEI may provide useful additional or improved information. The mouse lung images suggest that enhanced visualization of the bronchial tree would be beneficial in diseases such as bronchiectasis. Visualization of small subsegmental bronchi would aid the diagnosis of bronchiolitis obliterans. The visualization of individual acini and adjacent small subsegmental bronchi would be very beneficial in the diagnosis of interstitial lung disease. In these disease processes there is often traction bronchiectasis of small sub- segmental bronchi, and it is possible that this technique could enable the precise distribution of interstitial changes to be made. For example, it may be possible to determine whether the changes are subplural, based on the interlobular septal structures or centri-lobular structures. Analysis of such distribution features has been shown by high resolution CT to be of great diagnostic value. Visualization of small cystic air spaces is a possibility with this technique and this may aid the diagnosis of diseases showing such features such as fibrosing alveolitis, lymphangioliomyomatosis and Langerhan's cell histiocytosis. With reference to the mouse liver images, if such clear visualization of normal hepatic structures could be seen in man this would have a number of diagnostic possibilities. Visualization of mass lesions such as metastases could be possible due to distortion and displacement of adjacent vessels. Diffuse liver disease may also reproduce more generalized architectural disturbance within this intrahepatic branching structure. If at least one of the branching structures shown is to the biliary tree this technique may also have some value in delineating causes of biliary obstruction. The enhanced visualization of the edges of cortical bone seen in refraction images may aid diagnosis of fractures and may also be beneficial in the imaging of metabolic bone disease where there is subperiostial bone resorption such as hyperparathyroidism. The increased trabecular detail seen using refraction enhanced images may also be beneficial in the diagnosis of metabolic bone disease. Osteroporosis, Pagets disease and hyperparathyroidism affect the number and morphology of the trabeculae and diagnosis of such conditions may be enhanced by employment of refraction imaging.
|
Conclusions
|
|---|
Diffraction enhanced X-ray imaging may have promising applications as a medical imaging tool. It is capable of producing wide dynamic range images of biological systems, showing enhanced bone and soft tissue detail, at doses similar to or lower than those used in conventional absorption imaging. Characteristically, boundaries between different tissue types, where the refractive index changes rapidly, are displayed with high contrast, giving an edge enhanced appearance. Further studies are required to determine the precise benefits and areas of application in the medical imaging field.
|
Acknowledgments
|
|---|
The invaluable assistance of the Academic Unit of Cancer Studies, University Hospital Nottingham and in particular TM Morris, is gratefully acknowledged.
|
Footnotes
|
|---|
Work supported by Elettra award No. 2000166 and EU grants HPRI-CT-1999-00033 and HPRI-CT-1999-50008. 
Received for publication June 17, 2002.
Revision received November 22, 2002.
Accepted for publication February 6, 2003.
|
References
|
|---|
- Momose A, Takeda T, Itai Y. Blood vessels: depiction at phase-contrast X-ray imaging without contrast agents in the mouse and rat—feasibility study. Radiology 2000;217:593–6.[Abstract/Free Full Text]
- Momose A, Takeda T, Itai Y, Yoneyama A, Hirano K. Perspectives for medical applications of phase-contrast X-ray imaging. In: Ando M, Uyama C, editors. Medical applications of synchrotron radiation. Tokyo: Springer-Verlag, 1998;54–61.
- Momose A, Takeda T, Itai Y, Hirano K. Phase-contrast X-ray computed tomography for observing biological soft tissues. Nat Med 1996;2:473–5.[Medline]
- Wilkins SW, Gureyev TE, Gao D, Pogany A, Stevenson AW. Phase-contrast imaging using polychromatic hard X-rays. Nature 1996;384:335–8.[CrossRef]
- Ingal VN, Beliaevskaya EA. Imaging of biological objects in the plane-wave diffraction scheme. Il Nuovo Cimento 1997;19:553–60.
- Chapman D, Thomlinson W, Johnston RE, Washburn D, Pisano E, Gmür N, et al. Diffraction enhanced X-ray imaging. Phys Med Biol 1997;42:2015–25.[CrossRef][Medline]
- Zhong Z, Thomlinson W, Chapman D, Sayers D. Implementation of diffraction-enhanced imaging experiments at the NSLS and APS. Nucl Instr and Meth A 2000;450:556–67.[CrossRef]
- Arfelli F, Bonvicini V, Bravin A, Cantatore G, Dalla Palma L, Di Michiel M, et al. Mammography with synchrotron radiation: phase-detection techniques. Radiology 2000;215:286–93.[Abstract/Free Full Text]
- Pisano ET, Johnston RE, Chapman D, Geradts J, Iacocca MV, Livasy CA, et al. Human breast cancer specimens: diffraction-enhanced imaging with histologic correlation—improved conspicuity of lesion detail compared with digital radiography. Radiology 2000;214:895–901.[Abstract/Free Full Text]
- Thomlinson W, Chapman D, Zhong Z, Johnston RE, Sayers D. Diffraction enhanced X-ray imaging. In: Ando M, Uyama C, editors. Medical applications of synchrotron radiation. Tokyo: Springer-Verlag, 1998;72–6.
- Jacobs E, Watson SA, Morris TM, Robertson JFR. A new murine syngeneic model of breast cancer metastasis to bone expressing PTHrP. Clin Exp Metastasis 2000;17:780 abstract P122.
- Arfelli F, Bonvicini V, Bravin A, Cantatore G, Castelli E, Dalla Palma L, et al. Improvements in the field of radiological imaging at the SYRMEP beamline. In: Barber HB, Roehrig H, editors. Medical applications of penetrating radiation. 1999 July 18–23; Denver. Proc SPIE 3770;1999:2–12.
- Suzuki Y, Yagi N, Uesugi K. X-ray refraction-enhanced imaging and a method for phase retrieval for a simple object. J Synchrotron Rad 2002;9:160–5.[Medline]
- Lewis RA, Rogers KD, Hall CJ, Hufton AP, Evans S, Menk RH, et al. Diffraction enhanced imaging: improved contrast, lower dose x-ray imaging. In: Antonuk LE, Yaffe MJ, editors. Medical Imaging 2002: Physics of Medical Imaging. 2002 February 23–28; San Diego. Proc SPIE 2002;4682:286–97.
- Khounsary AM, MacDonald CA, editors. Advances in laboratory-based x-ray sources and optics II. 2001 July 29–August 3; San Diego. Proc SPIE 2001;4502.
- Kotre CJ, Birch IP. Phase contrast enhancement of X-ray mammography: a design study. Phys Med Biol 1999;44:2853–66.[Medline]
This article has been cited by other articles:
|
|
|
|
 
M. L. Siew, A. B. te Pas, M. J. Wallace, M. J. Kitchen, R. A. Lewis, A. Fouras, C. J. Morley, P. G. Davis, N. Yagi, K. Uesugi, et al.
Positive end-expiratory pressure enhances development of a functional residual capacity in preterm rabbits ventilated from birth
J Appl Physiol,
May 1, 2009;
106(5):
1487 - 1493.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M J Kitchen, R A Lewis, N Yagi, K Uesugi, D Paganin, S B Hooper, G Adams, S Jureczek, J Singh, C R Christensen, et al.
Phase contrast X-ray imaging of mice and rabbit lungs: a comparative study
Br. J. Radiol.,
November 1, 2005;
78(935):
1018 - 1027.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
