X-ray coherent scattering form factors of tissues, water and plastics using energy dispersion

Mechanical and Imaging Properties of a Clinical-Grade Kidney Phantom Based on Polydimethylsiloxane and Elastomer

Medical imaging phantoms are considered critical in mimicking the properties of human tissue for calibration, training, surgical planning, and simulation purposes. Hence, the stability and accuracy of the imaging phantom play a significant role in diagnostic imaging. This study aimed to evaluate the influence of hydrogen silicone (HS) and water (H₂O) on the compression strength, radiation attenuation properties, and computed tomography (CT) number of the blended Polydimethylsiloxane (PDMS) samples, and to verify the best material to simulate kidney tissue. Four samples with different compositions were studied, including samples S1, S2, S3, and S4, which consisted of PDMS 100%, HS/PDMS 20:80, H₂O/PDMS 20:80, and HS/H₂O/PDMS 20:40:40, respectively. The stability of the samples was assessed using compression testing, and the attenuation properties of sample S2 were evaluated. The effective atomic number of S2 showed a similar pattern to the human kidney tissue at 1.50 × 10⁻¹ to 1 MeV. With the use of a 120 kVp X-ray beam, the CT number quantified for S2, as well measured 40 HU, and had the highest contrast-to-noise ratio (CNR) value. Therefore, the S2 sample formulation exhibited the potential to mimic the human kidney, as it has a similar dynamic and is higher in terms of stability as a medical phantom.

Application of machine learning classifiers to X-ray diffraction imaging with medically relevant phantoms

PURPOSE

Recent studies have demonstrated the ability to rapidly produce large field of view X-ray diffraction (XRD) images, which provide rich new data relevant to the understanding and analysis of disease. However, work has only just begun on developing algorithms that maximize the performance toward decision-making and diagnostic tasks. In this study, we present the implementation of and comparison between rules-based and machine learning (ML) classifiers on XRD images of medically relevant phantoms to explore the potential for increased classification performance.

METHODS

Medically relevant phantoms were utilized to provide well-characterized ground-truths for comparing classifier performance. Water and polylactic acid (PLA) plastic were used as surrogates for cancerous and healthy tissue, respectively, and phantoms were created with varying levels of spatial complexity and biologically relevant features for quantitative testing of classifier performance. Our previously developed X-ray scanner was used to acquire co-registered X-ray transmission and diffraction images of the phantoms. For classification algorithms, we explored and compared two rules-based classifiers (cross-correlation, or matched-filter, and linear least-squares unmixing) and two ML classifiers (support vector machines and shallow neural networks). Reference XRD spectra (measured by a commercial diffractometer) were provided to the rules-based algorithms, while 60% of the measured XRD pixels were used for training of the ML algorithms. The area under the receiver operating characteristic curve (AUC) was used as a comparative metric between the classification algorithms, along with the accuracy performance at the midpoint threshold for each classifier.

RESULTS

The AUC values for material classification were 0.994 (cross-correlation [CC]), 0.994 (least-squares [LS]), 0.995 (support vector machine [SVM]), and 0.999 (shallow neural network [SNN]). Setting the classification threshold to the midpoint for each classifier resulted in accuracy values of CC = 96.48%, LS = 96.48%, SVM = 97.36%, and SNN = 98.94%. If only considering pixels ±3 mm from water-PLA boundaries (where partial volume effects could occur due to imaging resolution limits), the classification accuracies were CC = 89.32%, LS = 89.32%, SVM = 92.03%, and SNN = 96.79%, demonstrating an even larger improvement produced by the machine-learned algorithms in spatial regions critical for imaging tasks. Classification by transmission data alone produced an AUC of 0.773 and accuracy of 85.45%, well below the performance levels of any of the classifiers applied to XRD image data.

CONCLUSIONS

We demonstrated that ML-based classifiers outperformed rules-based approaches in terms of overall classification accuracy and improved the spatially resolved classification performance on XRD images of medical phantoms. In particular, the ML algorithms demonstrated considerably improved performance whenever multiple materials existed in a single voxel. The quantitative performance gains demonstrate an avenue to extract and harness XRD imaging data to improve material analysis for research, industrial, and clinical applications.

Accuracy of the independent atom approximation in digital tomosynthesis Monte Carlo simulations

Monte Carlo (MC) codes serve as the gold standard simulation tool during design and optimisation of x-ray imaging systems. Such simulations often model Rayleigh scattering based on the Independent Atom Approximation Model (IAM). This model neglects the low range molecular interference (MI) effects of non-crystalline materials such as human tissues. Previous work has found discrepancies in the simulated images of planar x-ray images between IAM and MI models. However, insignificant differences were found for computed tomography (CT) reconstructions. In this work we present Geant4 MC simulations of a flat panel source digital tomosynthesis (DT) system for human extremities. Results show that with a 1:9 scatter to primary ratio (SPR) in the x-ray projections, the DT reconstructions are insensitive to the differences of the IAM and MI models. Therefore, MC codes that use the IAM model are sufficient for the study of DT systems. That is because DT algorithms have a larger effect on image quality than the few percent change in the noise due to a physical model and noise suppression methods make this change even less important. Dependency of this conclusion on SPR must be considered in other DT modalities where SPR might be larger.

Identification of amyloid plaques in the brain using an x-ray photon-counting strip detector

Brain aggregates of β amyloid (βA) protein plaques have been widely recognized as associated with many neurodegenerative diseases, and their identification can assist in the early diagnosis of Alzheimer’s disease. We investigate the feasibility of using a spectral x-ray coherent scatter system with a silicon strip photon-counting detector for identifying brain βA protein plaques. This approach is based on differences in the structure of amyloid, white and grey matter in the brain. We simulated an energy- and angular-dispersive X-ray diffraction system with an x-ray pencil beam and Silicon strip sensor, energy-resolving detectors. The polychromatic beam is geometrically focused toward a region of interest in the brain. First, the open-source MC-GPU code for Monte Carlo transport was modified to accommodate the detector model. Second, brain phantoms with and without βA were simulated to assess the method and determine the radiation dose required to obtain acceptable statistical power. For βA targets of 3, 4 and 5 mm sizes in a 15-cm brain model, the required incident exposure was about 0.44 mR from a 60 kVp tungsten spectrum and 3.5 mm of added aluminum filtration. The results suggest that the proposed x-ray coherent scatter technique enables the use of high energy x-ray spectra and therefore has the potential to be used for accurate in vivo detection and quantification of βA in the brain within acceptable radiation dose levels.

X-ray diffraction tomography with limited projection information

X-ray diffraction tomography (XDT) records the spatially-resolved X-ray diffraction profile of an extended object. Compared to conventional transmission-based tomography, XDT displays high intrinsic contrast among materials of similar electron density and improves the accuracy in material identification thanks to the molecular structural information carried by diffracted photons. However, due to the weak diffraction signal, a tomographic scan covering the entire object typically requires a synchrotron facility to make the acquisition time more manageable. Imaging applications in medical and industrial settings usually do not require the examination of the entire object. Therefore, a diffraction tomography modality covering only the region of interest (ROI) and subsequent image reconstruction techniques with truncated projections are highly desirable. Here we propose a table-top diffraction tomography system that can resolve the spatially-variant diffraction form factor from internal regions within extended samples. We demonstrate that the interior reconstruction maintains the material contrast while reducing the imaging time by 6 folds. The presented method could accelerate the acquisition of XDT and be applied in portable imaging applications with a reduced radiation dose.

Non-invasive classification of breast microcalcifications using x-ray coherent scatter computed tomography

We investigate the use of energy dispersive x-ray coherent scatter computed tomography (ED-CSCT) as a non-invasive diagnostic method to differentiate between type I and type II breast calcifications. This approach is sensitive to the differences of composition and internal crystal structure of different types of microcalcifications. The study is carried out by simulating a CSCT system with a scanning pencil beam, considering a polychromatic x-ray source and an energy-resolving photon counting detector. In a first step, the multidimensional angle and energy distributed CSCT data is reduced to the projection-space distributions of only a few components, corresponding to the expected target composition: adipose, glandular tissue, weddellite (calcium oxalate) for type I calcifications, and hydroxyapatite for type II calcifications. The maximum-likelihood estimation of scatter components algorithm used, operating in the projection space, takes into account the polychromatic source, the detector response function and the energy dependent attenuation. In the second step, component images are reconstructed from the corresponding estimated component projections using filtered backprojection. In a preliminary step the coherent scatter differential cross sections for hydroxyapatite and weddellite minerals were determined experimentally. The classification of type I or II calcifications is done using the relative contrasts of their components as the criterion. Simulation tests were carried out for different doses and energy resolutions for multiple realizations. The results were analyzed using relative/receiver operating characteristic methodology and show good discrimination ability at medium and higher doses. The noninvasive CSCT technique shows potential to further improve the breast diagnostic accuracy and reduce the number of breast biopsies.

Snapshot fan beam coded aperture coherent scatter tomography

We use coherently scattered X-rays to measure the molecular composition of an object throughout its volume. We image a planar slice of the object in a single snapshot by illuminating it with a fan beam and placing a coded aperture between the object and the detectors. We characterize the system and demonstrate a resolution of 13 mm in range and 2 mm in cross-range and a fractional momentum transfer resolution of 15%. In addition, we show that this technique allows a 100x speedup compared to previously-studied pencil beam systems using the same components. Finally, by scanning an object through the beam, we image the full 4-dimensional data cube (3 spatial and 1 material dimension) for complete volumetric molecular imaging.

WAXS fat subtraction model to estimate differential linear scattering coefficients of fatless breast tissue: phantom materials evaluation

PURPOSE

Develop a method to subtract fat tissue contributions to wide-angle x-ray scatter (WAXS) signals of breast biopsies in order to estimate the differential linear scattering coefficients μ(s) of fatless tissue. Cancerous and fibroglandular tissue can then be compared independent of fat content. In this work phantom materials with known compositions were used to test the efficacy of the WAXS subtraction model.

METHODS

Each sample 5 mm in diameter and 5 mm thick was interrogated by a 50 kV 2.7 mm diameter beam for 3 min. A 25 mm(2) by 1 mm thick CdTe detector allowed measurements of a portion of the θ = 6° scattered field. A scatter technique provided means to estimate the incident spectrum N(0)(E) needed in the calculations of μ(s)[x(E, θ)] where x is the momentum transfer argument. Values of [Formula: see text] for composite phantoms consisting of three plastic layers were estimated and compared to the values obtained via the sum [Formula: see text], where ν(i) is the fractional volume of the ith plastic component. Water, polystyrene, and a volume mixture of 0.6 water + 0.4 polystyrene labelled as fibphan were chosen to mimic cancer, fat, and fibroglandular tissue, respectively. A WAXS subtraction model was used to remove the polystyrene signal from tissue composite phantoms so that the μ(s) of water and fibphan could be estimated. Although the composite samples were layered, simulations were performed to test the models under nonlayered conditions.

RESULTS

The well known μ(s) signal of water was reproduced effectively between 0.5 < x < 1.6 nm(-1). The [Formula: see text] obtained for the heterogeneous samples agreed with [Formula: see text]. Polystyrene signals were subtracted successfully from composite phantoms. The simulations validated the usefulness of the WAXS models for nonlayered biopsies.

CONCLUSIONS

The methodology to measure μ(s) of homogeneous samples was quantitatively accurate. Simple WAXS models predicted the probabilities for specific x-ray scattering to occur from heterogeneous biopsies. The fat subtraction model can allow μ(s) signals of breast cancer and fibroglandular tissue to be compared without the effects of fat provided there is an independent measurement of the fat volume fraction ν(f). Future work will consist of devising a quantitative x-ray digital imaging method to estimate ν(f) in ex vivo breast samples.

Snapshot molecular imaging using coded energy-sensitive detection

We demonstrate a technique for measuring the range-resolved coherent scatter form factors of different objects from a single snapshot. By illuminating the object with an x-ray pencil beam and placing a coded aperture in front of a linear array of energy-sensitive detector elements, we record the coherently scattered x-rays. This approach yields lateral, range, and momentum transfer resolutions of 1 mm, 5 mm, and 0.2 nm⁻¹, respectively, which is sufficient for the distinguishing a variety of solids and liquids. These results indicate a path toward real-time volumetric molecular imaging for non-destructive examination in a variety of applications, including medical diagnostics, quality inspection, and security detection.

Synchrotron-based coherent scatter x-ray projection imaging using an array of monoenergetic pencil beams

Traditional projection x-ray imaging utilizes only the information from the primary photons. Low-angle coherent scatter images can be acquired simultaneous to the primary images and provide additional information. In medical applications scatter imaging can improve x-ray contrast or reduce dose using information that is currently discarded in radiological images to augment the transmitted radiation information. Other applications include non-destructive testing and security. A system at the Canadian Light Source synchrotron was configured which utilizes multiple pencil beams (up to five) to create both primary and coherent scatter projection images, simultaneously. The sample was scanned through the beams using an automated step-and-shoot setup. Pixels were acquired in a hexagonal lattice to maximize packing efficiency. The typical pitch was between 1.0 and 1.6 mm. A Maximum Likelihood-Expectation Maximization-based iterative method was used to disentangle the overlapping information from the flat panel digital x-ray detector. The pixel value of the coherent scatter image was generated by integrating the radial profile (scatter intensity versus scattering angle) over an angular range. Different angular ranges maximize the contrast between different materials of interest. A five-beam primary and scatter image set (which had a pixel beam time of 990 ms and total scan time of 56 min) of a porcine phantom is included. For comparison a single-beam coherent scatter image of the same phantom is included. The muscle-fat contrast was 0.10 ± 0.01 and 1.16 ± 0.03 for the five-beam primary and scatter images, respectively. The air kerma was measured free in air using aluminum oxide optically stimulated luminescent dosimeters. The total area-averaged air kerma for the scan was measured to be 7.2 ± 0.4 cGy although due to difficulties in small-beam dosimetry this number could be inaccurate.