Retrospective gating on animal studies with microCT has gained popularity in recent years. Previously, we use ECG signals for cardiac gating and breathing airflow or video signals of abdominal motion for respiratory gating. This method is adequate and works well for most applications. However, through the years, researchers have noticed some pitfalls in the method. For example, the additional signal acquisition step may increase failure rate in practice. X-Ray image-based gating, on the other hand, does not require any extra step in the scanning. Therefore we investigate imagebased gating techniques. This paper presents a comparison study of the image-based versus signal-based approach to retrospective gating. The two application areas we have studied are respiratory and cardiac imaging for both rats and mice. Image-based respiratory gating on microCT is relatively straightforward and has been done by several other researchers and groups. This method retrieves an intensity curve of a region of interest (ROI) placed in the lung area on all projections. From scans on our systems based on step-and-shoot scanning mode, we confirm that this method is very effective. A detailed comparison between image-based and signal-based gating methods is given. For cardiac gating, breathing motion is not negligible and has to be dealt with. Another difficulty in cardiac gating is the relatively smaller amplitude of cardiac movements comparing to the respirational movements, and the higher heart rate. Higher heart rate requires high speed image acquisition. We have been working on our systems to improve the acquisition speed. A dual gating technique has been developed to achieve adequate cardiac imaging.
Image registration is a powerful tool in various tomographic applications. Our main focus is on microCT
applications in which samples/animals can be scanned multiple times under different conditions or at different
time points. For this purpose, a registration tool capable of handling fairly large volumes has been developed,
using a novel pseudo-3D method to achieve fast and interactive registration with simultaneous 3D visualization.
To reduce computation complexity in 3D registration, we decompose it into several 2D registrations, which are
applied to the orthogonal views (transaxial, sagittal and coronal) sequentially and iteratively. After registration
in each view, the next view is retrieved with the new transformation matrix for registration. This reduces the
computation complexity significantly. For rigid transform, we only need to search for 3 parameters (2 shifts, 1
rotation) in each of the 3 orthogonal views instead of 6 (3 shifts, 3 rotations) for full 3D volume. In addition, the
amount of voxels involved is also significantly reduced.
For the proposed pseudo-3D method, image-based registration is employed, with Sum of Square Difference
(SSD) as the similarity measure. The searching engine is Powell's conjugate direction method. In this paper,
only rigid transform is used. However, it can be extended to affine transform by adding scaling and possibly
shearing to the transform model. We have noticed that more information can be used in the 2D registration if
Maximum Intensity Projections (MIP) or Parallel Projections (PP) is used instead of the orthogonal views. Also,
other similarity measures, such as covariance or mutual information, can be easily incorporated.
The initial evaluation on microCT data shows very promising results. Two application examples are shown:
dental samples before and after treatment and structural changes in materials before and after compression.
Evaluation on registration accuracy between pseudo-3D method and true 3D method has been performed.
We have developed a compact grating-based in-vivo phase-contrast micro-CT system with a rotating gantry. The 50 W microfocus x-ray source operates with 20 to 50 kV peak energy. The length of the rotating interferometer is around 47 cm. Pixel size in the object is 30 micron; the field of view is approx. 35 mm in diameter, suited to image a mouse. The interferometer consists of three gratings: an absorption grating close to the x-ray source, a phase grating to introduce a π/2 phase shift and an absorption analyzer grating positioned at the first fractional Talbot distance. Numerous drives and actuators are used to provide angular and linear grating alignment, phase stepping and object/gantry precision positioning. Phantom studies were conducted to investigate performance, accuracy and stability of the scanner. In particular, the influences of gantry rotation and of temperature fluctuations on the interferometric image acquisition were characterized. Also dose measurements were performed. The first imaging results obtained with the system show the complementary nature of phase-contrast micro-CT images with respect to absorption-based micro-CT. Future improvements, necessary to optimize the scanner for in-vivo small-animal CT scanning on a regular and easy-to-use basis, are also discussed.
In our dual-modality microCT/microXRF system, the two sub-systems are combined in one machine, sharing a
travelling sample holder. The microXRF, based on a pin-hole collimator and a photon-counting energysensitive 2D-detector, obtains 3D chemical composition maps of a sample. These images often lack structural information. With the built-in microCT, 3D structural information of the sample can be obtained. The two subsystems need to be properly calibrated and aligned. This calibration and alignment procedure needs to be done for all pin-hole collimators, but only need to be performed once after the system is assembled. The two modalities are calibrated separately, by analyzing projection images of a 3-ball phantom. The phantom is made of a very thin plastic cylinder, on which 3 copper balls are attached at well-chosen locations. The same phantom is used for both sub-systems and is scanned sequentially. We have evaluated this calibration method on various CT scanners and it has proven to be very effective. But it is more challenging for the XRF subsystem due to the strong absorptions. The two imaging spaces are calibrated relative to their own coordinate systems. To align the two sub-systems, the centers of the balls in reconstructed volumes are determined and then aligned using a rigid transformation. Repeated tests have shown that the mechanical movements are stable and the reconstructed image volumes can
be well co-registered.
Proper selection of modern key components allows eliminating most artifacts in micro-CT and nano-CT systems already
during data acquisition. X-ray cameras with direct photon detection allow avoiding ring artifacts. Newly developed fully
depleted CCD sensors show an energy response similar to traditional cameras with a thin scintillator, but without any
geometrical distortions and flashes from x-ray photons penetrating through the fiber optics. Air-bearing rotation stages
and piezo-positioning minimizes mechanical inaccuracies in acquiring angular projections. Beam hardening can be
eliminated by energy-selective photon counting imaging.
X-ray fluorescence (XRF) allows imaging of the chemical composition of a specimen. We developed a 2<sup>nd</sup> generation
prototype laboratory system that can produce 3D chemical maps using microXRF as well as volumetric microCT
images. The latter can be used to overlay morphological information on top of the XRF image for co-registration. It is
also employed for attenuation correction during the tomographic reconstruction of the XRF images. The new system has
various hardware and software changes to improve the performance, stability and flexibility. A deep depleted CCD was
employed to improve the detection efficiency for high-energy fluorescence X rays. The use of a deep depleted CCD
requires signal-clustering techniques to correct for charge diffusion in the CCD to obtain the correct energy of the
fluorescence x rays. Furthermore, energy drift correction techniques were put in place to ensure stability of energy
measurement during very long scan times. To minimize the contribution of the long CCD readout times to the total scan
time, the exposure frames are dynamically adjust during the scan to the maximum time allowed for operation under
photon counting mode. The XRF component has a spatial resolution of 70 μm and an energy resolution of 180 eV at 6.4
An integrated microCT/microXRF system has been designed and built at SkyScan. The two sub-systems are aligned. The
microCT provides 3D morphological information of the sample, which can be also used for attenuation correction during
microXRF reconstruction. The microXRF, based on a pin-hole collimator and a photon-counting energy-sensitive 2Ddetector,
obtains 2D projections of 3D chemical composition inside the sample with 50-70 microns spatial resolution.
The reconstruction of 3D microXRF scans is challenging because of very low photon counting statistics due to limited
power of laboratory x-ray sources and the strong self-absorption of the low-energy fluorescence photons. We have
developed a maximum-likelihood expectation-maximization (ML-EM) algorithm based on Poisson model. This
algorithm has proven to be rather robust and good reconstructions have been obtained with sample scans. Regularization
is necessary to achieve stable reconstruction. One method is to apply smoothing between iterations. Two different
smoothing kernels have been evaluated: 3D symmetric Gaussian kernel and minimization of total variation. For further
improvement, a multi-ray resolution recovery technique has been evaluated.
The self-absorption is currently compensated by a simplified method: the correction coefficients are pre-calculated and
obtained by forward-projecting the attenuation map for both the primary X-rays and the fluorescence photons. The
attenuation maps at the energy of fluorescence photons are approximated from the CT image.
We have developed an x-ray computer tomography (CT) add-on to perform X-ray micro- and nanotomography in any
scanning electron microscope (SEM). The electron beam inside the SEM is focused on a metal target to generate x-rays.
Part of the X-rays pass through the object that is installed on a rotation stage. Shadow X-ray images are collected by a
CCD camera with direct photon detection mounted on the external wall of the SEM specimen chamber. An extensive
description on the working principles of this micro/nano-CT add-on together with some examples of CT-scans will be
given in this paper. The resolution that can be obtained with this set-up and the influence of the shape of the electron
beam are discussed. Furthermore, possible improvements on this SEM-CT set-up will be discussed: replacing the backilluminated
CCD with a fully depleted CCD with improved quantum efficiency (QE) for higher energies, reduces the
exposure time by 6 when using metal targets with x-ray characteristic lines around 10 keV.
Reconstruction theory requires that an object should be fully inside the field of view (FOV) of the scanning geometry. This implies that the number of pixels in the detector determines the smallest resolvable details for a given FOV size. Many commercially available micro-CT scanners use a 1-megapixel cameras with 1K pixels in horizontal direction, which can resolve features of 1/1000 of the FOV size. Using a large format detector and a few offset positions will
increase imaging resolution, but will also dramatically reduces
number of X-ray photons collected per pixel. To improve acquisition efficiency without compromising scanning time, we developed and implemented in commercially produced SkyScan-1172 scanners an adaptive geometry approach. To achieve a chosen magnification, the distances between x-ray source, detector and object are adjusted automatically to most compact geometry with maximum use of X-ray. This
adaptive geometry improves significantly the acquisition speed for a large range of magnifications and allows using 10+ megapixels detectors instead of detectors with 1-2 megapixels. Flexible acquisition geometry also opens possibility to use phase-contrast enhancement for improvement in spatial resolution.
Micro-CT and especially nano-CT scanning requires very high mechanical precision and stability of object manipulator, which is difficult to reach. Several other problems, such as drift of emission point inside an X-ray source, thermal expansion in different parts of the scanner, mechanical vibrations, and object movement or shrinkage during long scans, can also contribute to geometrical inaccuracies. All these inaccuracies result in artifacts which reduce achievable spatial resolution. Linear distortions can be partially compensated by rigid X/Y shifts in projection images. More complicated object movement and shrinkage will require non-linear transforms. This paper investigates techniques to compensate geometrical inaccuracies by linear transformation only. We have developed two methods to estimate individual X/Y shifts in each measured projection. The first method aligns measured projections with forward-projected projections
iteratively to reach an optimal X/Y shift estimation. It is more suitable for mechanical inaccuracies caused by random and jittery movement. The second method uses a very short reference scan acquired immediately after a main scan to obtain estimates of X/Y shifts. This method is rather effective for mechanical inaccuracies caused by slow and coherent mechanical drifts. Both methods have been implemented and evaluated on multiple scanners. Significant improvements
in image quality have been observed.
We have developed a compact laboratory scanner, which combines X-ray microtomography (microCT) with X-ray microfluorescence tomography (3D microXRF). This dual-modality scanner opens possibility for nondestructive threedimensional volumetric analysis of local chemical composition, enhanced by morphological information provided by the
built-in microCT. Unlike known microXRF methods based on collimated beam and detector, our microXRF scanner includes a full-field acquisition system based on an energy-sensitive detector with 512x512 pixels operating in photon counting mode. It allows detection of two-dimensional photon energy distribution in the range of 3...20keV. Up to 8 sets of energy windows can be selected for independent and simultaneous collection of microXRF images. By object rotation
the scanner acquires projections in transmission and fluorescence mode for subsequent 3D reconstruction. The system acquires data in such a way that the CT scans and XRF scans match each other in magnification and angular position. This makes image registration much easier and more accurate. MicroCT data is reconstructed with a FBP algorithm. All microXRF datasets are reconstructed by a maximum likelihood iterative algorithm, which uses corresponding CT images
for absorption correction.
High resolution micro-CT scanners are becoming widely used for <i>in vivo</i> imaging of small laboratory animals. However, imaging of chest area remains a challenging task due to periodic respiratory and cardiac motion, where respiratory motion dominates. To reduce motion artifacts and to allow dynamic imaging, we propose a retrospective
synchronization method for scans of chest area in our <i>in vivo</i> micro-CT scanners. In this synchronization method, we acquire projection images in a step-and-shoot mode, where multiple images are acquired covering more than one motion cycle at each step with exact time marks of every acquisition. In the meanwhile motion signals are recorded. An offline sorting program has been developed to sort images into corresponding motion phases. We have evaluated our method on respiratory motion. Compared to prospective synchronization methods, our method has several advantages: 1. flexible in
sorting; 2. continuous imaging: maximum utilization of radiation dose applied to the animal; 3. possibility for 4D dynamic imaging; 4. can be used during irregular breathing cycle. This method has been applied to two of SkyScan <i>in-vivo</i> scanners. Initial results indicate that the proposed method is adequate.