Xizeng Wu

Clinical implementation of x-ray phase-contrast imaging: theoretical foundations and design considerations.

Theoretical foundation and design considerations of a clinical feasible x-ray phase contrast imaging technique were presented in this paper. Different from the analysis of imaging phase object with weak absorption in literature, we proposed a new formalism for in-line phase-contrast imaging to analyze the effects of four clinically important factors on the phase contrast. These are the body parts attenuation, the spatial coherence of spherical waves from a finite-size focal spot, and polychromatic x-ray and radiation doses to patients for clinical applications. The theory presented in this paper can be applied widely in diagnostic x-ray imaging procedures. As an example, computer simulations were conducted and optimal design parameters were derived for clinical mammography. The results of phantom experiments were also presented which validated the theoretical analysis and computer simulations.

X-ray phase-attenuation duality and phase retrieval.

Phase retrieval is the key to quantitative x-ray phase-contrast imaging. To retrieve the phase image of an x-ray wave field, in general one needs multiple phase-contrast images. We have made a new observation of phase-attenuation duality for soft tissues, and we show how only a single phase-contrast image is needed for successful phase retrieval based on this duality. The phase-retrieval formula based on a single phase-contrast image of inhomogeneous soft tissue is derived and presented. We show the striking enhancement of the tissue contrast in simulated phase images that this new approach produces.

Molybdenum target x-ray spectra: a semiempirical model.

A semiempirical model for generating molybdenum target x-ray spectra is presented. The model is an extension of a previous model developed by the authors for tungsten and takes into account the depth of production for both bremsstrahlung and characteristic x-ray photons. As in the previous work, the optimal model parameters were determined using nonlinear least-squares fits to experimental data. Good agreement between the two was obtained. By varying target angle, off-axis angle, and filtration in the model in accordance with the x-ray tube and geometry of interest, results consistent with tabulated spectra for different conditions have been obtained.

A new theory of phase-contrast x-ray imaging based on Wigner distributions

There is a pressing need for a comprehensive theory for phase-contrast x-ray imaging to guide its development and clinical applications. This work presents such a theory as the foundation for deriving these guidelines. The new theory is based on the Wigner-distributions for the parabolic wave equations, and it is more general than the present theories based on the Fresnel-Kirchhoff diffraction theory. The new theory shows for the first time how the complex degree of coherence (CDC) of the incident x-ray beam determines the phase-contrast visibility in general, and how the reduced complex degree of coherence (RCDC) for an anode-source is equal to the systems optical transfer function for geometric unsharpness in particular. The role of detector resolution in phase visibility has been clarified as well. Computer simulations based on the new theory were conducted and optimal design parameters were derived for phase-contrast mammography systems.

Parametrization of mammography normalized average glandular dose tables

Data from existing tables of normalized glandular doses in mammography were parametrized to determine analytical expressions that match tabulated results within known uncertainties. The parametrization was performed for three different target/filter combinations (molybdenum target-molybdenum filter, molybdenum target-rhodium filter, and rhodium target-rhodium filter) and three different breast compositions (100% adipose, 50% adipose-50% glandular, and 100% glandular). The analytical expressions provide the normalized glandular dose for any breast composition within stated ranges of tabulated input parameters (kVp, half-value layer, and breast thickness). The maximum difference between tabulated and parametrized data is 1.29%, which is smaller than the stated 2% uncertainty in tabulated dose data due to uncertainties in the x-ray spectra models used to generate the tables. Analytical expressions are easily coded to create custom functions that return the normalized glandular dose for the set of input parameters. Examples of implementation are presented in Microsoft Visual Basic for Applications (VBA).

Clarification of aspects in in-line phase-sensitive x-ray imaging

In this report we clarify two aspects for in-line phase-sensitive x-ray imaging, which includes phase-contrast imaging and phase imaging. First, we point out that there is confusion in the literature about the lateral coherence length, which is widely adopted as the coherence criteria for implementing phase-sensitive imaging. The confusion exaggerates the coherence requirement for clinical implementation of in-line phase-sensitive imaging. Instead we show that the ratio of the phase-space shearing length to lateral coherence length is a good measure for gauging the partial coherence realized in a specific image setting. Second, based on the general intensity equation for in-line phase-sensitive imaging, we discuss the differences between the phase-contrast imaging and phase imaging in terms of the physics mechanism, image acquisition approaches, computation algorithm development, and the potentials for tissue quantitative characterization.

Phase retrieval from one single phase contrast x-ray image

Phase retrieval is required for achieving artifact-free x-ray phase-sensitive 3D imaging. A phase-retrieval approach based on the phase-attenuation duality with high energy x-rays can greatly facilitate for phase sensitive imaging by allowing phase retrieval from only one single projection image. The previously derived phase retrieval formula is valid only for small Fresnel propagator phases corresponding to common clinical imaging tasks. In this work we presented a new duality-based phase retrieval formula that can be applied for cases with large Fresnel propagator phases corresponding to high spatial resolution imaging. The computer simulation demonstrated superiority of this new formula over the previous phase retrieval formula in reconstructing the high frequency components of imaged objects. A modified Tikhonov regularization technique has been devised for phase retrieval in cases of very high resolution and large object-detector distance such that some Fresnel propagator phases may be close or greater than pi. This new phase retrieval formula lays the foundation for implementing high-resolution phase-sensitive 3D imaging of soft tissue objects.

An experimental method of determining relative phase-contrast factor for x-ray imaging systems

The relative phase-contrast factor (RPF) represents a quantitative measure of the phase-contrast transfer in x-ray in-line phase-contrast imaging. The larger the modulus of RPF(u, v) is, the more the phase-contrast manifests. In this work we show how the RPF can be determined by measurements of the focal spot size and x-ray spectra for a x-ray imaging system with a micro-focus x-ray tube. The results show the significant effects of x-ray beam hardening on the visibility of phase-contrast, and reveal a new dimension in seeking optimal techniques for x-ray phase-contrast imaging.

Preliminary Feasibility Study of an In-line Phase Contrast X-Ray Imaging Prototype

In this study, a series of imaging experiments on biological specimens, including human breast core biopsies, lumpectomy, and chicken tissues, as well as standard phantoms, were performed in an effort to investigate the feasibility of an in-line phase contrast X-ray imaging prototype. The prototype system employed in the study consists of a microfocus X-ray source with tungsten target and a digital flat panel detector, and it can be operated in both conventional attenuation-based imaging mode and in-line phase contrast imaging mode. Biological specimens were imaged in the conventional mode and phase contrast mode with the same source-to-image-detector distance (SID), and phase contrast images exhibited both improved image quality compared with conventional images, and the overshooting patterns along the boundaries in the specimens, which revealed the occurrence of the edge enhancement effect provided by the phase contrast technique. In addition, the performance of the phase contrast mode and conventional mode was compared based on the American College of Radiology (ACR) phantom imaging and contrast detail mammography (CDMAM) phantom-based contrast detail analysis with two experimental settings: one with the same SID and the other with the same object entrance exposure. In both pairs of comparison under our experimental conditions, the phase contrast imaging mode exhibited improved image quality as compared to the conventional mode, which further supported the feasibility of the prototype.

Three-dimensional x-ray fluorescence mapping of a gold nanoparticle-loaded phantom

PURPOSE X-ray fluorescence (XRF) is a promising technique with sufficient specificity and sensitivity for identifying and quantifying features in small samples containing high atomic number (Z) materials such as iodine, gadolinium, and gold. In this study, the feasibility of applying XRF to early breast cancer diagnosis and treatment is studied using a novel approach for three-dimensional (3D) x-ray fluorescence mapping (XFM) of gold nanoparticle (GNP)-loaded objects in a physical phantom at the technical level. METHODS All the theoretical analysis and experiments are conducted under the condition of using x-ray pencil beam and a compactly integrated x-ray spectrometer. The penetrability of the fluorescence x-rays from GNPs is first investigated by adopting a combination of BR12 with 70 mm/50 mm in thickness on the excitation/emission path to mimic the possible position of tumor goldin vivo. Then, a physical phantom made of BR12 is designed to translate in 3D space with three precise linear stages and subsequently the step by step XFM scanning is performed. The experimental technique named as background subtraction is applied to isolate the gold fluorescence from each spectrum obtained by the spectrometer. Afterwards, the attenuations of both the incident primary x-ray beam with energies beyond the gold K-edge energy (80.725 keV) and the isolated gold Kα fluorescence x-rays (65.99 -69.80 keV) acquired after background subtraction are well calibrated, and finally the unattenuated Kα fluorescence counts are used to realize mapping reconstruction and to describe the linear relationship between gold fluorescence counts and corresponding concentration of gold solutions. RESULTS The penetration results show that the goldKα fluorescence x-rays have sufficient penetrability for this phantom study, and the reconstructed mapping results indicate that both the spatial distribution and relative concentration of GNPs within the designed BR12 phantom can be well identified and quantified. CONCLUSIONS Although the XFM method in this investigation is still studied at the technical level and is not yet practical for routinein vivo mapping tasks with GNPs, the current penetrability measurements and phantom study strongly suggest the feasibility to establish and develop a 3D XFM system.