BJR
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
British Journal of Radiology (2005) 78, 1018-1027
© 2005 British Institute of Radiology
doi: 10.1259/bjr/13024611
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kitchen, M J
Right arrow Articles by PAvlov, K M
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kitchen, M J
Right arrow Articles by PAvlov, K M

Full Paper

Phase contrast X-ray imaging of mice and rabbit lungs: a comparative study

M J Kitchen, BSc(Hons)1, R A Lewis, PhD2, N Yagi, PhD3, K Uesugi, MSc3, D Paganin, PhD1, S B Hooper, PhD4, G Adams, PhD5, S Jureczek, PhD5, J Singh, PhD5, C R Christensen, PhD5, A P Hufton, PhD6, C J Hall, PhD7, K C Cheung, PhD7 and K M PAvlov, PhD1,2

1 School of Physics, Monash University, Victoria 3800, Australia, 2 Monash Centre for Synchrotron Science, Monash University, Victoria 3800, Australia, 3 SPring-8/JASRI, Mikazuki, Hyogo 679-5198, Japan, 4 Department of Physiology, Monash University, Victoria 3800, Australia, 5 University of Saskatchewan, Saskatoon, Saskatchewan S7N 0X4, Canada, 6 North Western Medical Physics, Christie Hospital, Manchester M20 4BX, UK and 7 Daresbury Laboratory, Warrington WA4 4AD, UK


   Abstract
Top
 Abstract
Introduction
 Phase contrast imaging
 Imaging methods
 Image analysis
 Results
 Discussion
 Conclusion
 References
 
The significant degree of X-ray phase contrast created by air-tissue interfaces, coupled with the poor radiographic contrast of conventional chest radiographs, makes the inflated lung an ideal candidate for investigating the potential diagnostic improvement afforded by phase contrast X-ray imaging. In small animals these methods highlight the lung airways and lobe boundaries and reveal the lung tissue as a speckled intensity pattern not seen in other soft tissues. We have compared analyser-based and propagation-based phase contrast imaging modalities, together with conventional radiographic imaging, to ascertain which technique shows the greatest image enhancement for various lung sizes. The conventional radiographic image of a mouse was obtained on a Siemens Nova 3000 mammography system, whilst phase contrast images of mice and rabbit chests were acquired at the medical imaging beamline (20B2) at the SPring-8 synchrotron radiation research facility in Japan. For mice aged 1 day, 1 week and 1 month old it was determined that analyser-based imaging showed the greatest overall image contrast, however, for an adult rabbit both techniques yielded excellent contrast. The success of these methods in creating high quality images for rabbit lungs raises the possibility of improving human lung imaging using phase contrast techniques.


   Introduction
Top
 Abstract
 Introduction
Phase contrast imaging
 Imaging methods
 Image analysis
 Results
 Discussion
 Conclusion
 References
 
Highly coherent X-ray sources allow one to image macroscopic biological structures using both synchrotron radiation [1–3] and laboratory sources [4–8], with much greater contrast than is possible using conventional radiographic systems. A full understanding of the physics underlying this process demands that X-rays be considered as waves rather than rays. In particular, the use of highly coherent X-rays (achieved with a small source size and/or large source-to-object distance, l1, and a narrow energy bandwidth [7]) can enhance image contrast since it is possible to detect the perturbations imposed by the object on the well-defined phase of the incident coherent wavefield (phase contrast) in addition to detecting the changes in beam attenuation (absorption contrast). In geometrical optics terms the phase changes cause transverse shifts in the local beam direction (refraction) upon propagation through media of varying refractive index, which can result in an intensity change in the image plane. These so-called "phase contrast imaging" techniques [1– 8] can significantly augment the visibility of interfaces between materials where refractive index changes occur. The air filled lung provides a unique biological structure with many such changes in refractive index, which make it an ideal tissue to study with these methods for diagnostic imaging purposes.

Detection of lung disease using conventional, absorption-based radiography is limited by the variation in tissue density created by a pathological process. Particularly in the early stages of lung disease, any density change (if present) is often too small to be detected by this imaging process. Even in their more advanced stages some diseases, particularly interstitial lung disease (ILD), can show completely normal (healthy) chest radiographs [9]. In fact, population studies have shown up to 100% of patients suffering from hypersensitivity pneumonitis (an ILD) appear to have normal chest radiographs [10]. Alternative methods of lung imaging include CT and MRI. Although CT techniques can improve the detection rate of such diseases by providing three-dimensional information, this method inherently exposes patients to relatively large X-ray doses compared with standard radiography [9]. MRI does not use ionizing radiation yet generally does not permit visualization of fine structures due to its relatively poor spatial resolution (typically around a millimetre) [9]. Phase contrast X-ray imaging techniques are therefore unique in providing high-resolution images at comparatively low dose [11].

The research objective of this study was to evaluate two phase contrast modalities for rodent lung imaging, comparing each with conventional radiography. These phase contrast modalities were analyser-based imaging (ABI) [3–6, 8, 11–14], sometimes called "diffraction enhanced imaging (DEI)" and propagation-based imaging (PBI) [1, 2, 7, 15, 16], alternatively called "refraction-enhanced imaging". Previous studies have compared these techniques for use in mammographic studies on a breast phantom and excised breast tissue samples [17, 18]. They showed both phase contrast techniques significantly increased breast image quality over absorption contrast radiography. It cannot be assumed that these results hold true for lung imaging since the lung's numerous air/tissue interfaces will perturb the X-ray wavefield in a different manner to breast tissue.

Both PBI and ABI have previously been used to study lungs of small mammals [13–16, 19], with both methods revealing a speckled intensity pattern across the lung which was not seen over other soft tissues. Suzuki et al [16] proposed that the speckle arises from multiple refraction of the X-ray beam through numerous air filled alveolar sacs. Computer simulations alongside PBI experiments on mice lungs support this theory, suggesting that the alveoli in projection act as aberrated compound refractive lenses [19].


   Phase contrast imaging
Top
 Abstract
 Introduction
 Phase contrast imaging
Imaging methods
 Image analysis
 Results
 Discussion
 Conclusion
 References
 
At X-ray energies every material has an energy dependent complex refractive indexGo


{1018equ1}

where {delta} is the refractive index decrement and {beta}={lambda}µl/4{pi} is the absorption index, µl is the linear attenuation coefficient and {lambda} is the X-ray wavelength. Radiography conventionally relies on {beta} alone (for absorption contrast), whereas phase contrast techniques are sensitive to both {delta} and {beta}. For biological tissue at diagnostic energies (10–100 keV) {delta} is typically around three to four orders of magnitude larger than {beta} [11]. As a result, phase contrast modalities can yield significantly greater image contrast over conventional radiography. Far from absorption edges, {delta} is inversely proportional to the square of the X-ray energy whilst {beta} decreases as the fourth power of energy [20]. The amount of phase contrast relative to absorption contrast therefore increases with X-ray energy. As a result there is potential to exploit phase contrast at higher X-ray energies than commonly used for absorption contrast, thereby reducing the dose [11].

Propagation-based phase contrast imaging (PBI)
PBI is a form of in-line holography [21], conventionally used with the propagation distance between sample and detector, l2, sufficiently small that Fresnel diffraction fringes do not obscure fine image detail, yet large enough to visualize refractive effects. In this setting, the geometrical optics approximation is generally applicable [16] such that image formation can be considered in terms of X-rays converging or diverging behind the object. This leads to a measurable refraction-induced intensity variation in the image [7, 22]. The effect is most prominent at object interfaces where phase changes are greatest (Figure 1Go). Biological tissues typically refract diagnostic energy X-rays on the micro radian scale, so the detection of such small deviations requires that l2 be sufficiently large, typically a few metres.



View larger version (12K):
[in this window]
[in a new window]
  		Figure 1. Propagation-based phase contrast imaging schematic showing the geometrical optics construction for imaging a cylindrical object.

 
PBI is largely dependent upon the incident X-ray wavefield having sufficient spatial coherence to ensure (for a quasi-monochromatic beam) the phase distribution of the wave across a sample is well defined. In geometrical optics terms, high spatial coherence provides well-defined ray paths from source to detector. When such a wavefield is perturbed by a sample it retains the phase information upon propagation to the image plane. For a polychromatic beam the situation is more complex although phase contrast can still be achieved [7]. Figure 1Go illustrates the propagation-based contrast mechanism and the experimental apparatus required for this technique.

PBI has previously been used by Suzuki et al [16] to image rat lungs in situ using 35 keV synchrotron radiation with l2=5.5 m, which provided excellent lung visibility, using a detector with 24 µm pixels having a spatial resolution of around 40 µm. They also imaged rabbit lungs in situ at 51 keV, with the higher energy being chosen to reduce attenuation, and l2 increased to 12 m to compensate for the decrease in refractive indices with increasing energy. For the rabbit they found the visibility of internal lung structures such as the bronchi to be very poor. They suggested that these structures were averaged out by the large number of objects (alveoli etc.) overlapping in projection. Our aims were to improve image quality with the use of a higher resolution detector (~25 µm), to compare mice and rabbit lung images at the same energy and propagation distance, and to contrast these results with the analyser-based technique.

Analyser-based phase contrast imaging (ABI)
For ABI a perfect crystal analyser is mounted between the object and detector (Figure 2Go). The crystal is aligned with the beam incident upon atomic planes where the Bragg condition is satisfied for a specific X-ray wavelength. Reflection from the analyser only occurs for X-rays within a narrow angular range of the Bragg reflection, with the resulting reflectivity versus incident angle being called the "rocking curve" (Figure 2Go inset). The reflected intensity after the analyser is determined not only by the object but also from the incident beam characteristics (e.g. X-ray energy and divergence) combined with the chosen crystallographic planes of both monochromator and analyser. Rays in the incident beam scattered outside the rocking curve (e.g. by the object) will produce negligible intensity at the detector, providing essentially scatter free images. In addition, this scatter rejection produces further image contrast where objects scatter radiation beyond the narrow width of the rocking curve [12].



View larger version (23K):
[in this window]
[in a new window]
  		Figure 2. Analyser-based phase contrast imaging schematic showing the geometrical optics construction for imaging a cylindrical object. The inset shows the analyser rocking curve for the Si(333) Bragg reflection as measured experimentally and calculated from theory (full width at half maximum ~5 x 10–6 radians). The theoretical curve incorporates the double-bounce Si(311) monochromator with the Si(333) analyser reflections and the known beam divergence for the beamline. {theta}L and {theta}H show the low and high angle points where the reflected intensity is at half of the peak intensity found at the Bragg angle {theta}B. Images are often acquired with the analyser tilted to either {theta}L or {theta}H since the intensity gradient with angle is a maximum, which provides large intensity changes for small angular deviations of X-rays.

 
Since radiation produced by a synchrotron's bending magnet is elongated in the horizontal direction, the reflecting planes of both monochromator and analyser crystal are therefore rotated with respect to the horizontal axis for alignment and provide a broad and intense reflected beam. This alignment establishes a vertical diffraction plane, meaning only phase changes in the vertical direction (of the incident beam) produce a change in image contrast. Figure 2Go describes how the intensity reflected by the analyser changes with incident X-ray angle, in terms of geometrical optics. The point-by-point image intensity is highly sensitive to the local beam direction (the phase gradient) with respect to the analyser's Bragg plane [11]. For a more complete description see Chapman et al [12].

Benefits of ABI include the filtering of scattered radiation and tuneable phase contrast via rotation of the analyser crystal. Its drawbacks include difficulties in aligning and maintaining the analyser position and its sensitivity to phase changes only in the diffraction plane.

In this paper the two phase contrast modalities, ABI and PBI, have been used to image mice and rabbit lungs in situ. For direct comparison we used the same animals for both methods on the one synchrotron beamline at the same X-ray energy. The aim was to determine which modality provided best lung tissue contrast, particularly for rabbit lungs that more closely resemble human lungs in size.

At this point it should be emphasised that synchrotron radiation has been used in this study, although both phase contrast techniques can be performed in the laboratory. In fact, it has been shown that polychromatic laboratory sources can be used for PBI and still provide excellent phase contrast [7, 23]. Furthermore, some of the earliest ABI experiments were performed on laboratory sources [4–6, 8]. Synchrotron radiation was used here primarily to provide a proof-of-principle by exploiting the high flux levels provided by the synchrotron facility, and to reduce image acquisition time while increasing signal-to-noise ratios.


   Imaging methods
Top
 Abstract
 Introduction
 Phase contrast imaging
 Imaging methods
Image analysis
 Results
 Discussion
 Conclusion
 References
 
To compare conventional radiography with the phase contrast techniques a Siemens Nova 3000 mammography system (Siemens, Erlangen, Germany) was used to image an adult mouse. The system was set at 26 kVp with a molybdenum target and filter at 7.2 mAs. The resultant image was scanned from film as a 16-bit digital image with 100 µm pixels.

Phase contrast imaging experiments were performed at the SPring-8 Synchrotron Radiation Research Institute, Japan, in Hutch 3 of beamline 20B2 in the Biomedical Imaging Station, approximately 210 m from the source point. Radiation from the bending magnet was made monochromatic by a double-bounce Si(311) monochromator tuned to an energy of 33 keV. Following previous studies [16], this energy was chosen to be high enough to provide sufficient transmission through both mice and rabbit chests, while still being low enough to retain good phase and absorption contrast. All animals were imaged with the ABI modality before removing the analyser to begin PBI on the same animals.

Images were recorded with a Hamamatsu phosphor charge-coupled device (CCD) detector (C4742-95HR; Hamamatsu, Japan) with a 24.0 mm x 15.7 mm active area and a measured spatial resolution of around 25 µm. Setting the detector in 2 x 2 binning mode provided a pixel size of approximately 12 µm. At the detector position, the beam was sufficiently large to cover the entire CCD field of view. Custom software was used to subtract the background detector offset, with the resulting image being normalized by dividing out the beam's intensity distribution having imaged the detector dark current and direct beam, respectively. Objects larger than the detector were imaged in segments and later tiled together digitally. Exposure times were limited by the detector saturation point for direct beam imaging, irrespective of the dose. For ABI the exposure times were further reduced since analyser drift and beam movements over time can produce noticeable contrast changes.

The ABI analyser was a 10 mm thick, 100 mm diameter Si(111) single crystal aligned close to the (333) Bragg reflection, at half the maximum reflectivity point, {theta}L. The inset to Figure 2Go shows contrast amplification is maximized at that point since the rocking curve gradient is steepest. Beam divergence, resulting partly from the dispersive setup between monochromator and analyser, necessitated the use of slits to reduce the beam to 11.3 mm x 0.7 mm for ABI only. Conventionally, fixed-source ABI (with a narrow beam) requires simultaneous scanning of both object and detector at matched speeds to prevent image blurring. Equipment limitations made this impractical, making it necessary to tile together multiple still images the size of the pre-analyser slits. Since this increased overall scan times, exposure times were further reduced for ABI.

PBI images were acquired at a propagation distance of 4.26 m as this was found to provide phase contrast fringes that (qualitatively) enhanced edge visibility of most structures, without diffraction effects obscuring fine details.

Approximate dose levels have been calculated for the phase contrast images based on the measured flux density of (3.4±0.1) x 107 photons s–1 mm–2 at 33 keV at the imaging position, for the maximum storage ring current of 100 mA (private communication, N Yagi, 2004). For each image this quantity has been re-scaled using the known current at the time of imaging. The in-air surface entrance dose for each animal was calculated for the given flux and exposure time using the known absorption coefficient of dry air at a given energy. For the conventional radiographic image the dose was calculated from the known tube settings and energy spectrum in the same manner.

Lung visibility was tested as a function of lung size, using phase contrast, by imaging lungs at various developmental stages. This included three male mice aged 1 day, 1 week, and 4 weeks old, respectively, and an adult male rabbit. Prior to imaging all animals were euthanized by an overdose of ether in accordance with regulations set by the SPring-8 Animal Care and Use Committee. One author (N Yagi) has found, using high-resolution CT studies on euthanized mice, that this method prevents severe pulmonary oedema, a side effect of numerous euthanasia methods. Live animals were not imaged in this study to avoid the motion artefacts of breathing. For ease of positioning during imaging, mice were suspended vertically from their tails and the rabbit was seated in an upright position. In addition, previous studies have shown that deceased lungs can maintain a sufficient aeration level to provide visibility of internal lung structure with phase contrast [13–16, 19].


   Image analysis
Top
 Abstract
 Introduction
 Phase contrast imaging
 Imaging methods
 Image analysis
Results
 Discussion
 Conclusion
 References
 
A quantitative assessment of image quality for the phase contrast techniques is difficult since contrast arises via different physical mechanisms. We use here two commonly known definitions of image quality, namely visibility and area contrast. In optics, the visibility V is often used to measure the contrast of interference fringes, which we here use to quantify the intensity variations seen as speckle. Visibility is defined asGo


{1018equ2}

where IMAX and IMIN are the maximum and minimum intensity values measured in the selected region of interest (ROI) [24]. To determine contrast between the lung and surrounding soft tissues we specify an area contrast, which relates the average intensity of a selected area, IAV, with that of a nearby region of relatively constant background intensity, IBG, by [25]Go


{1018equ3}

Black rectangular boxes in all figures outline the ROIs used for the lung in all calculations, with white rectangles showing the chosen region for the background intensity, IBG, used for area contrast measurements. All lung ROIs were positioned above the diaphragm over a region of relatively thick lung tissue covering a relatively large area to provide reasonable statistics. ROIs for the conventional image contained 400 (100 µm) pixels with phase contrast image ROIs ranging from 22 500 to 160 000 pixels. For the rabbit lung the ROI was positioned between the ribs, away from any visible features such as lobe boundaries. ROIs for IBG measures were chosen in a featureless region just below the lung.

Uncertainties in calculated quantities were estimated by repeating measurements on at least three ROIs of equivalent size positioned over portions of similar lung thickness for each animal, depending on the lung size with respect to the ROI dimensions.


   Results
Top
 Abstract
 Introduction
 Phase contrast imaging
 Imaging methods
 Image analysis
 Results
Discussion
 Conclusion
 References
 
Figure 3Go displays the conventional radiograph of an adult mouse on an 8-bit greyscale palette. Significant contrast is noticeable between bone and soft tissues, however the soft tissues are relatively indistinguishable from one another. Although the body position is not ideal there is no obvious lung structure visible. To measure contrast the greyscale values were normalized to the direct beam level with black as the maximum intensity. Although this erroneously assumes a linear relationship between greyscale and intensity, it is a reasonable premise for the narrow band of greyscales used in the selected regions of interest. Measures of both speckle visibility and area contrast for the ROIs in this image are compared in Figures 8Go and 9Go alongside those of the phase contrast images of Figures 4–7GoGoGoGo.



View larger version (67K):
[in this window]
[in a new window]
  		Figure 3. An adult mouse imaged using a Siemens Nova 3000 conventional mammography system at 26 kVp, 7.2 mAs. Image size: 87.8 mm x 46.7 mm. Black region of interest (ROI): 20 x 20 pixels. White ROI: 20 x 20 pixels. Surface entrance dose: 1.0±0.1 mGy.

 


View larger version (14K):
[in this window]
[in a new window]
  		Figure 8. Speckle visibility calculated for the images in Figures 3–7GoGoGoGoGo for the regions of interest (ROIs) denoted by the black rectangles.

 


View larger version (13K):
[in this window]
[in a new window]
  		Figure 9. Area contrast calculated for the images in Figures 3–7GoGoGoGoGo. Regions of interest (ROIs) denoted by black and white rectangles in these images define the object and background regions, respectively.

 


View larger version (27K):
[in this window]
[in a new window]
  		Figure 4. Phase contrast images of a 1-day-old male mouse imaged at 33 keV. (a) Propagation-based imaging (PBI) with l2=4.26 m. Image size: 34.61 mm x 13.26 mm. Black region of interest (ROI): 150 x 150 pixels. White ROI: 40 x 40 pixels. Exposure time: 5.0 s. Surface entrance dose: 8.6±0.3 mGy. (b) Analyser-based imaging (ABI) image of the same mouse. Image size: 45.66 mm x 11.10 mm. Black ROI: 150 x 150 pixels. White ROI: 40 x 40 pixels. Exposure time: 0.5 s. Surface entrance dose: 0.92±0.03 mGy.

 


View larger version (19K):
[in this window]
[in a new window]
  		Figure 5. Phase contrast images of a 1-week-old male mouse imaged at 33 keV. (a) Propagation-based imaging (PBI) with l2=4.26 m. Image size: 54.07 mm x 16.79 mm. Black region of interest (ROI): 200 x 200 pixels. White ROI: 70 x 70 pixels. Exposure time: 5.0 s. Surface entrance dose: 8.6±0.3 mGy. (b) Analyser-based imaging (ABI) image of the same mouse. Image size: 64.26 mm x 11.29 mm. Black ROI: 200 x 200 pixels. White ROI: 50 x 50 pixels. Exposure time: 0.5 s. Surface entrance dose: 0.95±0.03 mGy.

 


View larger version (146K):
[in this window]
[in a new window]
  		Figure 6. Phase contrast images of a 1-month-old male mouse imaged at 33 keV. (a) Propagation-based imaging (PBI) with l2=4.26 m. Image size: 95.93 mm x 32.22 mm. Black region of interest (ROI): 300 x 300 pixels. White ROI: 100 x 100 pixels. Exposure time: 5.0 s. Surface entrance dose: 8.6±0.3 mGy. (b) Analyser-based imaging (ABI) image of the same mouse. Image size: 47.16 mm x 20.00 mm. Black ROI: 300 x 300 pixels. White ROI: 100 x 100 pixels. Exposure time: 0.5 s. Surface entrance dose: 0.91±0.03 mGy. (c, d) Magnified segments of the lung from (a) and (b), respectively.

 


View larger version (73K):
[in this window]
[in a new window]
  		Figure 7. Phase contrast images of an adult male rabbit imaged at 33 keV. (a) Propagation-based imaging (PBI) with l2=4.26 m. Image size: 63.31 mm x 77.42 mm. Black region of interest (ROI): 400 x 400 pixels. White ROI: 250 x 250 pixels. Exposure time: 5.0 s. Surface entrance dose: 9.8±0.3 mGy. (b) Analyser-based imaging (ABI) of the same rabbit. Image size: 57.74 mm x 60.86 mm. Black ROI: 400 x 400 pixels. White ROI: 250 x 250 pixels. Exposure time: 3.0 s. Surface entrance dose: 6.2±0.3 mGy.

 
Figures 4–6GoGoGo show phase contrast images of three male mice at various ages for both phase contrast modalities. Each image has been displayed using an 8-bit greyscale with the palette normalized to the minimum and maximum values observed within the black ROIs. All images are displayed as "negatives" for comparison with the conventional radiograph. Larger intensity values therefore appear darker than lower intensity values.

Immediately apparent in the PBI image of the 1-day-old mouse (Figure 4aGo), in comparison with the conventional image, are the sinuses, airways and digestive gases in the intestinal tract. The lung tissue is also highly visible as an area of marked intensity variations, or "speckles", similar to that noted by previous workers in the field [13, 15, 16]. These features are also apparent in the ABI image of the same mouse shown in Figure 4bGo. The ventral portion of the diaphragm is visible in both images, although more prominently in the ABI case, as a curved line overlying the basal lobes of both lungs.

Images of the 1-week and 1-month-old mice are displayed in Figures 5 and 6GoGo, respectively. Features seen in the 1-day-old mouse are again seen for these mice, although often with greater clarity due to increased size and tissue development (e.g. increased bone density). Importantly, the speckled intensity pattern remains evident across the lungs making them stand out above other organs. Figure 7Go shows phase contrast images of an adult male rabbit chest. The lung tissue appears highly speckled for both modalities in these images, highlighting it against surrounding tissues. No bronchi are visible in either rabbit image, confirming the results of Suzuki et al [16], although the lung boundaries can be distinguished and there is evidence of a lobe boundary below the fourth rib of the left lung in Figure 7aGo.

Figure 8Go shows the calculated visibility of the speckles (Equation 2Go) with Figure 9Go illustrating the overall lung area contrast (Equation 3Go) for absorption and phase contrast images of the mice and rabbit shown in Figures 3–7GoGoGoGoGo. For each of the ROIs a histogram of intensities was plotted, with each showing a roughly Gaussian intensity distribution. To avoid statistical outliers affecting IMAX and IMIN values (Equation 2Go), the maximum and minimum values were taken at three standard deviations from the mean histogram intensity. Variability of photon statistics between images was considered since exposure times were not constant. To remove such variability, intensity histograms of the direct beam were computed for each image. The standard deviations of these "background" distributions were subtracted in quadrature from the standard deviations of the appropriate lung speckle distributions. Here it was assumed that intensity fluctuations within the lung speckle and the background intensity were statistically independent.

The surface entrance dose received by the adult mouse imaged with the conventional system was 1.0±0.1 mGy. All three mice imaged with PBI were exposed for 5 s (per image segment) with each receiving a relatively large dose of 8.6±0.3 mGy. The rabbit was exposed for the same duration at a larger ring current and received a dose of 9.8±0.3 mGy. Using ABI each mouse was exposed for just 0.5 s per section (for stability reasons mentioned previously) at slightly different ring currents, providing doses of 0.92±0.03 mGy, 0.95±0.03 mGy and 0.91±0.03 mGy for the 1-day-, 1-week- and 1-month-old mice, respectively. The rabbit received a 6.2±0.3 mGy dose per 3 s exposure for this technique.


   Discussion
Top
 Abstract
 Introduction
 Phase contrast imaging
 Imaging methods
 Image analysis
 Results
 Discussion
Conclusion
 References
 
Technique comparison
Quantitative comparison between the conventional and synchrotron based images in Figures 3–7GoGoGoGoGo shows the increase in lung visibility with phase contrast. However, the improvement is not due to phase contrast effects alone. Excellent coherence of the synchrotron radiation alongside high detector spatial resolution provides a marked improvement in image contrast over the conventional approach (see for example references [11] and [15]).

With phase contrast the lung tissue clearly stands out from surrounding soft tissues, even for rabbit lungs. From that perspective phase contrast modalities may prove valuable for diagnostic lung imaging. Improved visibility of the lung results from the speckled intensity pattern. When the lung becomes sufficiently thick this speckle can reduce visibility of features such as bronchi, suggesting that the phase contrast should perhaps be reduced by altering experimental parameters (e.g. propagation distance) to recover such features. The inability to detect these features, as also noted by Suzuki et al [16], is evident in Figure 7Go despite the high resolution used here.

Although speckle can reduce visibility of structures along the same ray paths as the lung, it may provide information regarding tissue structure that could be decoded to reveal, for example, the mean size and distribution of the alveoli themselves, which would in turn reveal information regarding lung health. We have begun researching these possibilities [19], although further modelling and experimentation on lung tissue is required to test this hypothesis. Future research will also involve studying the effects of lung diseases on these speckle patterns.

Evaluation of current results
Contrast measures shown in Figure 8Go have considerable error margins since the speckle visibility varies across the tissue. The results are unlikely to be exactly reproducible using an alternative imaging system with a different animal, and merely indicate a trend that may be expected for similar experiments using these techniques. It must be noted that contrast values will vary as a function of propagation distance, X-ray energy and the rocking curve reflection.

Under the experimental parameters used here, Figures 8 and 9GoGo suggest that ABI provides the best image quality for mice lung tissue in terms of both area contrast and speckle visibility. The lack of phase contrast and low spatial resolution in the conventional image is responsible for the narrow intensity histogram indicative of poor speckle visibility. Not surprisingly, however, the area contrast is similar for the conventional image and the mice imaged using propagation based phase contrast, since the latter approach simply redistributes the intensity without changing the net transmitted intensity. For the larger rabbit lungs both techniques show excellent, yet similar, speckle visibility and area contrast. Figure 9Go suggests the area contrast is best for the PBI method, although it should be noted that this quantity is highly dependent on the chosen regions of interest. Using the same energy and propagation distance (33 keV at 4.26 m) the rabbit lung surprisingly showed better overall contrast than all three mice. We conclude that good contrast can be found for rabbit lungs under the appropriate conditions, which suggests applicability of phase contrast to human lung imaging.

Despite the scatter rejecting properties of the analyser, for rabbit lung images the ABI modality showed effectively no improvement in overall image contrast compared with those obtained using the PBI approach. It also showed no improvement in the visibility of features such as bronchi. The relatively low rabbit lung area contrast may have resulted from analyser and/or beam drift, and possibly from the upper limit placed on IMAX by the rocking curve amplitude. The effects of analyser drift are noticeable in the rabbit lung image of Figure 7bGo as relatively sharp intensity changes between tiled segments. For the ABI images this contrast may be readily changed, and perhaps improved, by rocking the analyser to a different position [13]. Despite the intensity gradient being largest (with respect to angle) at {theta}L, the contrast is not necessarily optimized for lung tissue at this angle. The tuneable contrast in ABI, as a parameter with respect to which the obtained images can be optimized, is a worthy subject for future lung imaging studies.

Effects of lung anatomy on phase contrast images
Unlike those of many mammals, alveoli in mice lungs primarily develop postnatally. In fact, at birth neonatal mice have few, if any, alveoli as their lungs predominantly consist of relatively simple primary saccules from which the alveoli form [26]. Effectively this means that the number of air filled cavities not only increases with chest size, but also depends on the developmental stage of the lung. Mature mouse alveoli range in diameter from 38 µm to 80 µm [26, 27] whereas those of mature rabbits are somewhat larger at 100–150 µm [28]. Saccules of the immature mice are more comparable in size with the alveoli of adult rabbits than of adult mice.

Despite this change in lung structure with age, the speckle visibility and area contrast for the mice remained effectively constant with the PBI method. (Whilst the intensity histogram remains similar, the speckle visibility does not describe the spatial distribution of the speckles. Analysis of the power spectra of these speckles could be used to reveal such information.) With terminal airways (saccules or alveoli) behaving in projection as aberrated compound refractive lenses, the appropriate propagation distance, l2, for maximum speckle contrast is highly dependent on the number and size of the terminal airways [19]. Evidently at this beam energy and propagation distance the rabbit lung structure was able to provide excellent image contrast.

Figure 8Go shows that speckle visibility decreases with increasing mouse age for the ABI images. The smaller, more numerous terminal airways (alveoli) of the oldest mouse must, on average, cause more scattering (by refraction) of the X-ray beam than the younger mice lungs. Each pixel then detects rays scattered from numerous angles, with the intensity of each ray depending on the rocking curve reflectivity for that angle (Figure 2Go). The recorded intensity is the sum of the contribution from these rays. We speculate that the intensity distribution across the lung narrows with increased scattering as each pixel records a larger spread of ray angles, causing an averaging of the integrated intensities.

For both phase contrast modalities the speckle visibility was significantly greater for the rabbit than for the mice (Figure 8Go). The 1-month-old mouse lungs were approximately 11.6 mm thick (as measured from a lateral view image) which gives a maximum of around 200 alveoli in projection. The rabbit lung thickness was estimated at 45 mm thick, giving a maximum of roughly 360 alveoli in projection. Both speckle visibility and area contrast for the rabbit were high, despite it possibly having almost twice as many alveoli in projection than the mouse. This likely results from the rabbit alveoli being significantly larger than the pixel dimensions, thereby reducing the averaging effects seen with the 1-month-old mouse lungs. Human lung imaging may also yield high speckle visibility under these conditions as human alveoli are larger again at 200–250 µm in diameter [26].

An interesting feature of the mouse lung anatomy is clearly revealed in Figures 4–6GoGoGo, particularly with the propagation technique, in the demarcation of the lung boundaries. Such imagery clearly illustrates the interaction between the lung and the chest wall and is particularly evident in Figure 6c,dGo. The chest wall plays an important role in opposing lung recoil and prevents the lung from collapsing at end-expiration; the high degree of recoil is predominantly caused by surface tension within the lung, despite the presence of surfactant. In Figure 6cGo the structural role of each rib in supporting lung expansion is clearly evident.

Imaging practicalities
For these experiments the energy, propagation distance and analyser reflection were selected to obtain high quality images of rodent lung tissue. For a given lung size such parameters should be optimized for maximum image contrast for the particular X-ray source. The choice of phase contrast modality depends upon the apparatus available and the object being studied. However, the simplicity of the PBI method generally makes it the more feasible option, as it can be performed on polychromatic laboratory sources [7], although currently with limited flux and therefore longer exposure times.

One of ABI's greatest deficiencies is its strong bias towards structural fluctuations that vary in a direction parallel to the plane of diffraction of the analyser crystal. This bias is clearly seen in the airways of the mice in Figures 4–6GoGoGo, where the vertically aligned trachea is almost invisible with ABI though it is quite obvious using PBI. This places some limitations on the technique. Conversely, the non-vertical bronchi are more visible with ABI as this technique is so sensitive to phase gradients.

Efforts to overcome this problem include the unification of ABI with PBI by simply increasing the propagation distance between object and detector in the ABI setup [30]. This hybrid method could prove useful in phase contrast lung imaging. ABI's lack of sensitivity to phase gradients in a particular direction may be used to advantage, to remove unwanted objects by aligning them in the diffraction plane. For example, vertically aligned hair covering the mice and rabbit thorax is virtually invisible in all the ABI images.


   Conclusion
Top
 Abstract
 Introduction
 Phase contrast imaging
 Imaging methods
 Image analysis
 Results
 Discussion
 Conclusion
References
 
Both PBI and ABI are capable of producing remarkable visibility of small animal lung tissue in comparison with conventional radiography. The enhanced lung visibility results from a speckled intensity pattern across the lung tissue and a highlighting of the lung boundaries. The ability to see such boundaries provides a non-invasive method of studying lung function.

Measuring image quality between various techniques is non-trivial for lung tissue since no two lungs will provide identical images, and there are many adjustable parameters between imaging modalities. This experiment suggests that for imaging lungs of small mammals, like mice, ABI can yield the greatest overall lung contrast against surrounding tissues and the largest visibility of the speckled intensity pattern itself. In this experiment the simpler propagation-based technique provided the greatest enhancement to contrast for the larger rabbit lungs, although the uncertainty is high with both techniques. It is plausible that either technique could also provide excellent image quality for larger lungs, showing a potential method of enhancing imaging methods for human lungs. Further analysis of the speckled intensity patterns seen with phase contrast could potentially assist in determining lung tissue health, though this research avenue is yet to be explored.

Whilst the role of phase contrast lung imaging in a clinical setting remains unclear it currently provides numerous benefits for non-invasively studying lung function and anatomy, at least for small animals, and may prove equally useful on larger sized lungs.


   Acknowledgments
 
We acknowledge the Access to Major Research Facilities Program (managed by the Australian Nuclear Science and Technology Organization) for supporting this work and funding the overseas visits of the Australian coauthors to conduct the experiments. We also thank the Japan Synchrotron Radiation Research Institute for the privilege of using the SPring-8 facility. M Kitchen wishes to acknowledge the support provided by an Australian Postgraduate Award. D Paganin and K Paulov acknowledge financial support from the Australian Research Council. We thank Dr T Sera of SPring-8/JASRI for his help with the animal experiments.


   Footnotes
 
Work supported by the Commonwealth of Australia under the Access to Major Research Facilities Program (AMRFP), proposal 02/03-S-15 and the Japan Synchrotron Radiation Research Institute (JASRI), proposal 2003A0181-NL2-np. Back

Received for publication February 25, 2005. Revision received April 27, 2005. Accepted for publication May 25, 2005.


   References
Top
 Abstract
 Introduction
 Phase contrast imaging
 Imaging methods
 Image analysis
 Results
 Discussion
 Conclusion
 References
 

  1. Snigirev A, Snigireva I, Kohn V, Kuznetsov S, Schelokov I. On the possibilities of x-ray phase contrast microimaging by coherent high-energy synchrotron radiation. Rev Sci Instrum 1995;66:5486–92.[CrossRef]
  2. Cloetens P, Barrett R, Baruchel J, Guigay J-P, Schlenker M. Phase objects in synchrotron radiation hard X-ray imaging. J Phys D Appl Phys 1996;29:133–46.[CrossRef]
  3. 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.
  4. Förster E, Goetz K, Zaumseil P. Double crystal diffractometry for the characterization of targets for laserfusion experiments. 1980;15:937–45.
  5. Somenkov VA, Tkalich AK, Shil'shtein SS. Refraction contrast in X-ray introscopy. Sov Phys-Tech Phys 1991;61:197–201.
  6. Davis TJ, Gao D, Gureyev TE, Stevenson AW, Wilkins SW. Phase-contrast imaging of weakly absorbing materials using hard X-rays. Nature 1995;373:595–8.[CrossRef]
  7. Wilkins SW, Gureyev TE, Gao D, Pogany A, Stevenson AW. Phase-contrast imaging using polychromatic hard X-rays. Nature 1996;384:335–8.[CrossRef]
  8. Ingal VN, Beliaevskaya EA. Imaging of biological objects in the plane-wave diffraction scheme. Il Nuovo Cimento 1997;19:553–60.
  9. Lynch DA, Newell JD, Lee JS. Imaging of diffuse lung disease. New York: B.C. Decker, 2000.
  10. Hodgson MJ, Parkinson DK, Karpf M. Chest X-rays in hypersensitivity pneumonitis: meta-analysis of a secular trend. Am J Ind Med 1989;16:45–53.[Medline]
  11. Lewis RA, Rogers KD, Hall CJ, Hufton AP, Evans S, Menk R, et al. Diffraction-enhanced imaging: improved contrast and lower dose X-ray imaging. In: Antonuk LE, Yaffe MJ, editors. Medical imaging 2002: physics of medical imaging. San Diego, CA: SPIE, 2002;268–97.
  12. 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]
  13. Lewis RA, Hall CJ, Hufton AP, Evans S, Menk RH, Arfelli F, et al. X-ray refraction effects: application to the imaging of biological tissues. Br J Radiol 2003;76:301–8.[Abstract/Free Full Text]
  14. 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]
  15. Yagi N, Suzuki Y, Umetani K, Kohmura Y, Yamasaki K. Refraction-enhanced x-ray imaging of mouse lung using synchrotron radiation source. Med Phys 1999;26:2190–3.[CrossRef][Medline]
  16. Suzuki Y, Yagi N, Uesugi K. X-ray refraction-enhanced imaging and a simple method for phase retrieval for a simple object. J Synchrotron Rad 2002;9:160–5.[CrossRef][Medline]
  17. Fiedler S, Pagot E, Cloetens P, Bravin A, Baruchel J, Härtwig J, et al. Evaluation of two phase contrast techniques: diffraction enhanced imaging and propagation. Proc SPIE 2003;5030:266–73.[CrossRef]
  18. Pagot E, Fiedler S, Cloetens P, Bravin A, Coan P, Fezzaa K, et al. Quantitative comparison between two phase contrast techniques: diffraction enhanced imaging and phase propagation imaging. Phys Med Biol 2005;50:709–24.[Medline]
  19. Kitchen MJ, Paganin D, Lewis RA, Yagi N, Uesugi K, Mudie ST. On the origin of speckle in x-ray phase contrast images of lung tissue. Phys Med Biol 2004;49:4335–48.[CrossRef][Medline]
  20. Gureyev TE, Mayo S, Wilkins SW, Paganin D, Stevenson AW. Quantitative in-line phase-contrast imaging with multienergy X rays. Phys Rev Lett 2001;86:5827–30.[Medline]
  21. Gabor D. A new microscopic principle. Nature 1948;161:777–8.[Medline]
  22. Nugent KA, Paganin D, Gureyev TE. A phase odyssey. Physics Today August 2001:27–32.
  23. Pogany A, Gao D, Wilkins SW. Contrast and resolution in imaging with a microfocus x-ray source. Rev Sci Instrum 1997;68:2774–82.[CrossRef]
  24. Hecht E. Optics (4th edn). Reading, MA: Addison Wesley Longman, Inc., 2002:602.
  25. Johns HE, Cunningham JR. The physics of radiology. Illinois: Charles C Thomas, 1983:570.
  26. Foster HL, Small JD, Fox JG. The mouse in biomedical research. In: Foster HL, Small JD, Fox JG, editors. Normative biology, immunology, and husbandry. New York: Academic Press, 1983:252–3.
  27. Irvin CG, Bates JHT. Measuring the lung function in the mouse: the challenge of size. Respir Res 2003;4:4.[CrossRef][Medline]
  28. Kovar J, Sly PD, Willet KE. Postnatal alveolar development of the rabbit. J Appl Physiol 2002;93:629–35.[Abstract/Free Full Text]
  29. Pavlov KM, Gureyev TE, Paganin D, Nesterets YI, Morgan MJ, Lewis RA. Linear systems with slowly varying transfer functions and their application to x-ray phase-contrast imaging. J Phys D Appl Phys 2004;37:2746–50.[CrossRef]



This article has been cited by other articles:


Home page
 J. Appl. Physiol.Home page
M. L. Siew, M. J. Wallace, M. J. Kitchen, R. A. Lewis, A. Fouras, A. B. te Pas, N. Yagi, K. Uesugi, K. K. W. Siu, and S. B. Hooper
Inspiration regulates the rate and temporal pattern of lung liquid clearance and lung aeration at birth
J Appl Physiol, June 1, 2009; 106(6): 1888 - 1895.
[Abstract] [Full Text] [PDF]

Home page
 J. Appl. Physiol.Home page
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]

Home page
 FASEB J.Home page
S. B. Hooper, M. J. Kitchen, M. J. Wallace, N. Yagi, K. Uesugi, M. J. Morgan, C. Hall, K. K. W. Siu, I. M. Williams, M. Siew, et al.
Imaging lung aeration and lung liquid clearance at birth
FASEB J, October 1, 2007; 21(12): 3329 - 3337.
[Abstract] [Full Text] [PDF]

Home page
 Br. J. Radiol.Home page
BJR Review of the Year - 2005.
Br. J. Radiol., March 1, 2006; 79(939): 183 - 187.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kitchen, M J
Right arrow Articles by PAvlov, K M
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kitchen, M J
Right arrow Articles by PAvlov, K M

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
BJR DMFR IMAGING  ALL BIR JOURNALS