X-ray transmission imaging has been used in a variety of applications for high-resolution measurements based on shape and density. Similarly, X-ray diffraction (XRD) imaging has been used widely for molecular structure-based identification of materials. Combining these X-ray methods has the potential to provide high-resolution material identification, exceeding the capabilities of either modality alone. However, XRD imaging methods have been limited in application by their long measurement times and poor spatial resolution, which has generally precluded combined, rapid measurements of X-ray transmission and diffraction. In this work, we present a novel X-ray fan beam coded aperture transmission and diffraction imaging system, developed using commercially available components, for rapid and accurate non-destructive imaging of industrial and biomedical specimens. The imaging system uses a 160 kV Bremsstrahlung X-ray source while achieving a spatial resolution of ≈ 1 × 1 mm2 and a spectral accuracy of > 95% with only 15 s exposures per 150 mm fan beam slice. Applications of this technology are reported in geological imaging, pharmaceutical inspection, and medical diagnosis. The performance of the imaging system indicates improved material differentiation relative to transmission imaging alone at scan times suitable for a variety of industrial and biomedical applications.

Compared to eye plaque brachytherapy, the proposed focused kV x-ray technique is noninvasive and shows great advantage in sparing healthy critical organs without sacrificing the tumor control. The NP radiation dose enhancement is considerable at our proposed kV range even with a low NP concentration in the tumor, providing better critical structure protection and more flexibility for treatment planning.

X-ray diffraction (XRD) imaging yields spatially resolved, material-specific information, which can aid medical diagnosis and inform treatment. In this work we used simulations to analyze the utility of fan beam coded aperture XRD imaging for fast, high-resolution scatter imaging of biospecimens for tissue assessment. To evaluate the proposed system’s utility in a specific task, we employed a deterministic model to produce simulated data from biologically realistic breast tissue phantoms and model-based reconstruction to recover a spatial map of the XRD signatures throughout the phantoms. We found an XRD spatial resolution of ≈1 mm with a mean reconstructed spectral accuracy of 0.98 ± 0.01 for a simulated 1 × 150 mm2 fan beam operating at 160 kVp, 10 mA, and 4.5 s exposures. A classifier for cancer detection was developed utilizing cross-correlation of XRD spectra against a spectral library, with a receiver operating characteristic curve with an area under the curve value of 0.972. Our results indicated a potential diagnostic modality that could aid in tasks ranging from analysis of ex-vivo pathology biospecimens to intraoperative cancer margin assessment, motivating future work to develop an experimental system while enabling the development of improved algorithms for imaging and tissue analysis-based classification performance.

Special attention is required in planning and administering radiation therapy to patients with cardiac implantable electronic devices (CIEDs), such as pacemaker and defibrillator. The range of dose to CIEDs that can induce malfunction is large among CIEDs. Clinically significant defects have been reported at dose as low as 0.15 Gy. Therefore, accurate estimation of dose to CIED and dose reduction are both important even if the dose is expected to be less than the often‐used 2‐Gy limit. We investigated the use of bolus in in vivo dosimetry for CIEDs. Solid water phantom measurements of out‐of‐field dose for a 6‐MV beam were performed using parallel plate chamber with and without 1‐ to 2‐cm bolus covering the chamber. In vivo dosimetry at skin surface above the CIED was performed with and without bolus covering the CIED for three patients with the CIED <5 cm from the field edge. Chamber measured dose at depth ~0.5–1.5 cm below the skin surface, where the CIED is normally located, was reduced by ~7–48% with bolus. The dose reduction became smaller at deeper depths and with smaller field size. In vivo dosimetry at skin surface also indicated ~20%–60% lower dose when using bolus for the three patients. The dose measured with bolus more accurately reflects the dose to CIED and is less affected by contaminant electrons and linac head scatter. In general, the treatment planning system (TPS) calculation underestimated the dose to CIED, but it predicts the CIED dose more accurately when bolus is used. We recommend the use of 1‐ to 2‐cm bolus to cover the CIED during in vivo CIED dose measurements for more accurate CIED dose estimation. If the CIED is placed <2 cm in depth and its dose is mainly from anterior beams, we recommend using the bolus during the entire course of radiation delivery to reduce the dose to CIED.

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 wellcharacterized 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 (crosscorrelation [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 rulesbased 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 532

No abstract

No abstract