Synchrotron-based computed microtomography (CMT) is an extension of conventional medical computed axial tomography (CAT) scans to high spatial resolution. By employing a nondestructive and in situ three-dimensional (3-D) imaging technique such as CMT, one can measure the internal structure or elemental distribution in a virtual cross-section of a biological specimen with- out the need for tissue preparation or sectioning that might cause loss of struc- tural integrity or redistribution of the elements. Three-dimensional images and virtual cross-sections of the X-ray attenuation or individual elemental fluorescence within the specimen can be rendered, allowing visualization of the variability in microstructure or elemental distribution.
For detailed information on the principles, data collection and process- ing of microtomography data, including discussion of different types of CMT (e.g. transmission, edge, fluorescence, and diffraction), see Sutton et al.
(2002). While X-ray diffraction techniques are not discussed in this review, examples of diffraction microtomography (DMT) can be found in Bleuet et al. (2008) and Lanzirotti et al. (2010). In recent years, there has been a growing interest in the use of phase-contrast imaging and phase-contrast tomography to obtain structural information, particularly for biological specimens which have low absorption-contrast yet decent phase-contrast because the X-ray phase-shift cross-section for light elements (e.g. C, H, O, N) is nearly a thousand times larger than their X-ray absorption cross- section (Momose et al., 1996). For additional information on phase-contrast imaging with hard X-rays, the reader is directed to Hornberger et al. (2008) and Holzner et al. (2010). The latter two references demonstrate the inte- gration of phase-contrast detection with scanning XRF microprobe, thus detailed structural information can be collected simultaneously with àXRF images or fluorescence computed microtomography (fCMT) sinograms.
The present discussion, however, will focus on transmission, edge and fCMT.
When elemental abundance is low, fCMT will be the method of choice. Synchrotron-based fCMT requires no sample pretreatment, allows
noninvasive examination of living materials, and can detect elemental abun- dances in the submicrogram per gram range with a resolution of 10 àm or less (Sutton et al., 2002). To use fCMT effectively, the absorption of emis- sion lines of interest must be low enough through the sample to allow for their detection given the object’s diameter (recall discussion from Section 2 concerning estimates for the fraction of X-rays transmitted through cellu- lose “leaf ” material of 300 àm thickness). Thus, absorption effects generally make this imaging technique most applicable to samples of small diameter (typically <2–3 mm), the analysis of higher energy emission lines and/or materials of low density (e.g. biological tissues); however, some degree of correction for absorption effects can be made (Schroer, 2001).
In the fCMT measurement, the sample is translated and rotated under a microfocused beam while the fluorescence intensities for multiple ele- ments are recorded with a solid-state detector and absorption contrast is recorded with a photodiode or ion chamber downstream of the sample.
The conventional microprobe apparatus is used with the addition of a rotation stage upon whose axis the object is centered with a goniometer head. Once centered, fluorescence is measured as the object is translated through the X-ray beam. In effect, the resultant [X, theta] arrays of data can then be reconstructed as [X,Y] virtual slices through the object including “air” pixels on each side. At the end of each translation, the sample is rotated slightly, and the measurement is repeated in this man- ner until the sample has traversed 180°. The optimal number of projec- tions (each line scan is a projection) is determined by the Nyquist limit for discrete sampling, and is Nπ/2, where N is the number of pixels per line (Dowd et al., 1999). The translation step size is chosen to be com- parable to the beam size. Thus, high-resolution fCMT requires many projections to yield voxels (3-D pixels) at the target resolution; however, it is somewhat common practice to undersample in the interest of time, and Sutton et al. (2002) suggest a rotational step size (in radians) equal to π/N, which sets the number of projections needed for 180° rotation equal to the number of pixels per line. The raw dataset consists of a posi- tion-versus-angle image (or sinogram) for each emission line monitored.
The resultant [X, theta] arrays of data can then be reconstructed as [X,Y]
virtual slices through the object using backprojection or fast Fourier transform reconstruction algorithms. While, in some cases, it would be impractical to image a significant volume of sample using fCMT, it can be used to generate a full three-dimensional dataset (i.e. slice-by-slice), as done by Kim et al. (2006) (Fig. 1.6).
fCMT measurements require typically several hours to complete a single slice, thus the specimen must withstand a substantial X-ray dose without changing shape or dehydrating. Often, it is necessary to dry or freeze-dry plant material for fCMT (seeds are an exception). An early application of fCMT to plant biology was reported by Hansel et al. (2001, 2002) who used fCMT in combination with àXRF and XAS to study Fe plaques and associated metals on the roots of reed canary grass and cattail collected from a wetland receiving drainage from a century-old Ag mine. A similar study by Blute et al. (2004) used fCMT to record oxidation-state tomograms of As(III) and As(V) on root plaque of cattail. For detailed explanations of
Figure 1.6 fCMT of Arabidopsis seed. (A) Light micrograph cross-section of a mature seed. (B, C) Total X-ray absorption tomographic slices of Columbia-0 (wild-type) and vit1-1 mutant seeds. (D) àXRF tomographic slices and composite images of Fe (blue), Mn (green) and Zn (red) Kα fluorescence lines from Columbia-0 and vit1-1. (E, F) Three- dimensional rendering of total X-ray absorption of a wild-type Arabidopsis seed. (G, H) Three-dimensional rendering of Fe Kα fluorescence in Columbia-0 and vit1-1, respec- tively. (Reprinted with permission from Kim et al. (2006)). See the color plate.
oxidation-state àXRF mapping and tomography, readers are directed to Lanzirotti et al. (2010), Marcus (2010), and Sutton et al. (1995, 2002). The basic approach is to make multiple àXRF maps or fCMT sinograms of the specimen, where the monochromatic energy for each image is chosen to preferentially excite particular oxidation-state components of the element of interest (including an image at the edge-step for normalization), and then the distributions of individual oxidation states are determined by deconvo- lution of these images. Thus, the specimen receives three or more times the radiation dose than from a single measurement. It is worth mentioning that both of the root plaque studies were conducted with freeze-dried tissues, and the latter study acknowledged that wet roots dried and lost structural integrity during the required analysis time. However, recent advances in XRF detectors and fast detector electronics that allow scanning a sample
“on-the-fly” have greatly reduced the overhead associated with sample positioning and detector readout, and these recent improvements have dra- matically reduced the exposure time for fCMT and àXRF measurements.
When elemental abundance is high (circa >1 wt%), one can use a full- field mode of CMT for element-specific imaging. For full-field CMT, the sample is exposed to a wide-fan X-ray beam, and one measures the transmit- ted X-rays that are converted to visible light via a single-crystal scintillator and then projected with a microscope objective onto an area detector. Since a wide-fan X-ray beam is used, the sample is only rotated in the X-ray beam (i.e. no translation). Advantages are that sample dimensions can be large, the sample can be imaged in minutes and does not need to withstand a large radiation dose, and thus hydrated samples can be analyzed; however, one does not directly obtain element-specific images using this method (only structural information). For element sensitivity, it is necessary to record the image both above and below the absorption edge of the ele- ment of interest and then subtract the images, which is called edge or dif- ferential absorption computed microtomography (DA-CMT). While the subtracted image offers element specificity, the below-edge image provides a glimpse of the internal structure of the specimen, and can be useful to overlay with the elemental image to explore spatial associations. Tappero et al. (2007) used DA-CMT to image the internal distribution of Co in hydrated Alyssum murale leaves, and by registering the elemental and struc- tural images, observed a distinct distribution of Co between cells (apoplastic) in the ground tissue and the localization of Co on the leaf surface near the tips and margins. Another advantage of the full-field mode for CMT is that more than a single slice is captured in the measurement, thus a full 3-D
volume of sample is recorded. For instance, the field of view in the vertical direction can be several millimeters with a pixel resolution on the order of 5 àm, thus a single measurement yields more than 500 tomographic slices.
These slices can be arranged into a movie sequence to view the changes in elemental distribution as one traverses the sample in the X, Y, or Z direc- tion. Additionally, these slices can be used to generate a full 3-D volume of the specimen that can be explored from every possible position and angle.