An indirect estimation of x-ray spectrum via convolutional neural network and transmission measurement

Objective.In this work, we aim to propose an accurate and robust spectrum estimation method by synergistically combining x-ray imaging physics with a convolutional neural network (CNN).Approach.The approach relies on transmission measurements, and the estimated spectrum is formulated as a convolutional summation of a few model spectra generated using Monte Carlo simulation. The difference between the actual and estimated projections is utilized as the loss function to train the network. We contrasted this approach with the weighted sums of model spectra approach previously proposed. Comprehensive studies were performed to demonstrate the robustness and accuracy of the proposed approach in various scenarios.Main results.The results show the desirable accuracy of the CNN-based method for spectrum estimation. The ME and NRMSE were -0.021 keV and 3.04% for 80 kVp, and 0.006 keV and 4.44% for 100 kVp, superior to the previous approach. The robustness test and experimental study also demonstrated superior performances. The CNN-based approach yielded remarkably consistent results in phantoms with various material combinations, and the CNN-based approach was robust concerning spectrum generators and calibration phantoms.Significance. We proposed a method for estimating the real spectrum by integrating a deep learning model with real imaging physics. The results demonstrated that this method was accurate and robust in estimating the spectrum, and it is potentially helpful for broad x-ray imaging tasks.

X-ray energy spectrum estimation based on a virtual computed tomography system

This paper presents a method for estimating the x-ray energy spectrum for computed tomography (CT) in the diagnostic energy range from the reconstructed CT image itself. To this end, a virtual CT system was developed, and datasets, including CT images for the Gammex phantom labeled by the corresponding energy spectra, were generated. Using these datasets, an artificial neural network (ANN) model was trained to reproduce the energy spectrum from the CT values in the Gammex inserts. In the actual application, an aluminum-based bow-tie filter was used in the virtual CT system, and an ANN model with a bow-tie filter was also developed. Both ANN models without/with a bow-tie filter can estimate the x-ray spectrum within the agreement, which is defined as one minus the absolute error, of more than 80% on average. The agreement increases as the tube voltage increases. The estimation was occasionally inaccurate when the amount of noise on the CT image was considerable. Image quality with a signal-to-noise ratio of more than 10 for the basis material of the Gammex phantom was required to predict the spectrum accurately. Based on the experimental data acquired from Activion16 (Canon Medical System, Japan), the ANN model with a bow-tie filter produced a reasonable energy spectrum by simultaneous optimization of the shape of the bow-tie filter. The present method requires a CT image for the Gammex phantom only, and no special setup, thus it is expected to be readily applied in clinical applications, such as beam hardening reduction, CT dose management, and material decomposition, all of which require exact information on the x-ray energy spectrum.

CT beam dosimetric characterization procedure for personalized dosimetry

Personalized dosimetry in computed tomography (CT) can be realized by a full Monte Carlo (MC) simulation of the scan procedure. Essential input data needed for the simulation are appropriate CT x-ray source models and a model of the patient's body which is based on the CT image. The purpose of this work is to develop comprehensive procedures for the determination of CT x-ray source models and their verification by comparison of calculated and measured dose distributions in physical phantoms. Mobile equipment together with customized software was developed and used for non-invasive determination of equivalent source models of CT scanners under clinical conditions. Standard and physical anthropomorphic CT dose phantoms equipped with real-time CT dose probes at five representative positions were scanned. The accumulated dose was measured during the scan at the five positions. ImpactMC, an MC-based CT dose software program, was used to simulate the scan. The necessary inputs were obtained from the scan parameters, from the equivalent source models and from the material-segmented CT images of the phantoms. 3D dose distributions in the phantoms were simulated and the dose values calculated at the five positions inside the phantom were compared to measured dose values. Initial results were obtained by means of a General Electric Optima CT 660 and a Toshiba (Canon) Aquilion ONE. In general, the measured and calculated dose values were within relative uncertainties that had been estimated to be less than 10%. The procedures developed were found to be viable and rapid. The procedures are applicable to any scanner type under clinical conditions without making use of the service mode with stationary x-ray tube position. Results show that the procedures are well suited for determining and verifying the equivalent source models needed for personalized CT dosimetry based on post-scan MC calculations.

Segmentation-free x-ray energy spectrum estimation for computed tomography using dual-energy material decomposition

An x-ray energy spectrum plays an essential role in computed tomography (CT) imaging and related tasks. Because of the high photon flux of clinical CT scanners, most of the spectrum estimation methods are indirect and usually suffer from various limitations. In this study, we aim to provide a segmentation-free, indirect transmission measurement-based energy spectrum estimation method using dual-energy material decomposition. The general principle of this method is to minimize the quadratic error between the polychromatic forward projection and the raw projection to calibrate a set of unknown weights, which are used to express the unknown spectrum together with a set of model spectra. The polychromatic forward projection is performed using material-specific images, which are obtained using dual-energy material decomposition. The algorithm was evaluated using numerical simulations, experimental phantom data, and realistic patient data. The results show that the estimated spectrum matches the reference spectrum quite well and the method is robust. Extensive studies suggest that the method provides an accurate estimate of the CT spectrum without dedicated physical phantom and prolonged workflow. This paper may be attractive for CT dose calculation, artifacts reduction, polychromatic image reconstruction, and other spectrum-involved CT applications.

A model-based scatter artifacts correction for cone beam CT

PURPOSE

Due to the increased axial coverage of multislice computed tomography (CT) and the introduction of flat detectors, the size of x-ray illumination fields has grown dramatically, causing an increase in scatter radiation. For CT imaging, scatter is a significant issue that introduces shading artifact, streaks, as well as reduced contrast and Hounsfield Units (HU) accuracy. The purpose of this work is to provide a fast and accurate scatter artifacts correction algorithm for cone beam CT (CBCT) imaging.

METHODS

The method starts with an estimation of coarse scatter profiles for a set of CBCT data in either image domain or projection domain. A denoising algorithm designed specifically for Poisson signals is then applied to derive the final scatter distribution. Qualitative and quantitative evaluations using thorax and abdomen phantoms with Monte Carlo (MC) simulations, experimental Catphan phantom data, and in vivo human data acquired for a clinical image guided radiation therapy were performed. Scatter correction in both projection domain and image domain was conducted and the influences of segmentation method, mismatched attenuation coefficients, and spectrum model as well as parameter selection were also investigated.

RESULTS

Results show that the proposed algorithm can significantly reduce scatter artifacts and recover the correct HU in either projection domain or image domain. For the MC thorax phantom study, four-components segmentation yields the best results, while the results of three-components segmentation are still acceptable. The parameters (iteration number K and weight β) affect the accuracy of the scatter correction and the results get improved as K and β increase. It was found that variations in attenuation coefficient accuracies only slightly impact the performance of the proposed processing. For the Catphan phantom data, the mean value over all pixels in the residual image is reduced from -21.8 to -0.2 HU and 0.7 HU for projection domain and image domain, respectively. The contrast of the in vivo human images is greatly improved after correction.

CONCLUSIONS

The software-based technique has a number of advantages, such as high computational efficiency and accuracy, and the capability of performing scatter correction without modifying the clinical workflow (i.e., no extra scan/measurement data are needed) or modifying the imaging hardware. When implemented practically, this should improve the accuracy of CBCT image quantitation and significantly impact CBCT-based interventional procedures and adaptive radiation therapy.

Relative electron density determination using a physics based parameterization of photon interactions in medical DECT

Radiotherapy and particle therapy treatment planning require accurate knowledge of the electron density and elemental composition of the tissues in the beam path to predict the local dose deposition. We describe a method for the analysis of dual energy computed tomography (DECT) images that provides the electron densities and effective atomic numbers of tissues. The CT measurement process is modelled by system weighting functions, which apply an energy dependent weighting to the parameterization of the total cross section for photon interactions with matter. This detailed parameterization is based on the theoretical analysis of Jackson and Hawkes and deviates, at most, 0.3% from the tabulated NIST values for the elements H to Zn. To account for beam hardening in the object as present in the CT image we implemented an iterative process employing a local weighting function, derived from the method proposed by Heismann and Balda. With this method effective atomic numbers between 1 and 30 can be determined. The method has been experimentally validated on a commercially available tissue characterization phantom with 16 inserts made of tissue substitutes and aluminium that has been scanned on a dual source CT system with tube potentials of 100 kV and 140 kV using a clinical scan protocol. Relative electron densities of all tissue substitutes have been determined with accuracy better than 1%. The presented DECT analysis method thus provides high accuracy electron densities and effective atomic numbers for radiotherapy and especially particle therapy treatment planning.