High-Density Glass Scintillators for Proton Radiography-Relative Luminosity, Proton Response, and Spatial Resolution.

Proton radiography is a promising development in proton therapy, and researchers are currently exploring optimal detector materials to construct proton radiography detector arrays. High-density glass scintillators may improve integrating-mode proton radiography detectors by increasing spatial resolution and decreasing detector thickness. We evaluated several new scintillators, activated with europium or terbium, with proton response measurements and Monte Carlo simulations, characterizing relative luminosity, ionization quenching, and proton radiograph spatial resolution. We applied a correction based on Birks's analytical model for ionization quenching. The data demonstrate increased relative luminosity with increased activation element concentration, and higher relative luminosity for samples activated with europium. An increased glass density enables more compact detector geometries and higher spatial resolution. These findings suggest that a tungsten and gadolinium oxide-based glass activated with 4% europium is an ideal scintillator for testing in a full-size proton radiography detector.

Integrated-mode proton radiography with 2D lateral projections.

Integrated-mode proton radiography leading to water equivalent thickness (WET) maps is an avenue of interest for motion management, patient positioning, andin vivorange verification. Radiographs can be obtained using a pencil beam scanning setup with a large 3D monolithic scintillator coupled with optical cameras. Established reconstruction methods either (1) involve a camera at the distal end of the scintillator, or (2) use a lateral view camera as a range telescope. Both approaches lead to limited image quality. The purpose of this work is to propose a third, novel reconstruction framework that exploits the 2D information provided by two lateral view cameras, to improve image quality achievable using lateral views. The three methods are first compared in a simulated Geant4 Monte Carlo framework using an extended cardiac torso (XCAT) phantom and a slanted edge. The proposed method with 2D lateral views is also compared with the range telescope approach using experimental data acquired with a plastic volumetric scintillator. Scanned phantoms include a Las Vegas (contrast), 9 tissue-substitute inserts (WET accuracy), and a paediatric head phantom. Resolution increases from 0.24 (distal) to 0.33 lp mm-1(proposed method) on the simulated slanted edge phantom, and the mean absolute error on WET maps of the XCAT phantom is reduced from 3.4 to 2.7 mm with the same methods. Experimental data from the proposed 2D lateral views indicate a 36% increase in contrast relative to the range telescope method. High WET accuracy is obtained, with a mean absolute error of 0.4 mm over 9 inserts. Results are presented for various pencil beam spacing ranging from 2 to 6 mm. This work illustrates that high quality proton radiographs can be obtained with clinical beam settings and the proposed reconstruction framework with 2D lateral views, with potential applications in adaptive proton therapy.

Focus stacking single-event particle radiography for high spatial resolution images and 3D feature localization.

Objective.We demonstrate a novel focus stacking technique to improve spatial resolution of single-event particle radiography (pRad), and exploit its potential for 3D feature detection.Approach.Focus stacking, used typically in optical photography and microscopy, is a technique to combine multiple images with different focal depths into a single super-resolution image. Each pixel in the final image is chosen from the image with the largest gradient at that pixel's position. pRad data can be reconstructed at different depths in the patient based on an estimate of each particle's trajectory (called distance-driven binning; DDB). For a given feature, there is a depth of reconstruction for which the spatial resolution of DDB is maximal. Focus stacking can hence be applied to a series of DDB images reconstructed from a single pRad acquisition for different depths, yielding both a high-resolution projection and information on the features' radiological depth at the same time. We demonstrate this technique with Geant4 simulated pRads of a water phantom (20 cm thick) with five bone cube inserts at different depths (1 × 1 × 1 cm3) and a lung cancer patient.Main results.For proton radiography of the cube phantom, focus stacking achieved a median resolution improvement of 136% compared to a state-of-the-art maximum likelihood pRad reconstruction algorithm and a median of 28% compared to DDB where the reconstruction depth was the center of each cube. For the lung patient, resolution was visually improved, without loss in accuracy. The focus stacking method also enabled to estimate the depth of the cubes within few millimeters accuracy, except for one shallow cube, where the depth was underestimated by 2.5 cm.Significance.Focus stacking utilizes the inherent 3D information encoded in pRad by the particle's scattering, overcoming current spatial resolution limits. It further opens possibilities for 3D feature localization. Therefore, focus stacking holds great potential for future pRad applications.

A likelihood-based particle imaging filter using prior information.

BACKGROUND: Particle imaging can increase precision in proton and ion therapy. Interactions with nuclei in the imaged object increase image noise and reduce image quality, especially for multinucleon ions that can fragment, such as helium. PURPOSE: This work proposes a particle imaging filter, referred to as the Prior Filter, based on using prior information in the form of an estimated relative stopping power (RSP) map and the principles of electromagnetic interaction, to identify particles that have undergone nuclear interaction. The particles identified as having undergone nuclear interactions are then excluded from the image reconstruction, reducing the image noise. METHODS: The Prior Filter uses Fermi-Eyges scattering and Tschalär straggling theories to determine the likelihood that a particle only interacts electromagnetically. A threshold is then set to reject those particles with a low likelihood. The filter was evaluated and compared with a filter that estimates this likelihood based on the measured distribution of energy and scattering angle within pixels, commonly implemented as the 3σ filter. Reconstructed radiographs from simulated data of a 20-cm water cylinder and an anthropomorphic chest phantom were generated with both protons and helium ions to assess the effect of the filters on noise reduction. The simulation also allowed assessment of secondary particle removal through the particle histories. Experimental data were acquired of the Catphan CTP 404 Sensitometry phantom using the U.S. proton CT (pCT) collaboration prototype scanner. The proton and helium images were filtered with both the prior filtering method and a state-of-the-art method including an implementation of the 3σ filter. For both cases, a dE-E telescope filter, designed for this type of detector, was also applied. RESULTS: The proton radiographs showed a small reduction in noise (1 mm of water-equivalent thickness [WET]) but a larger reduction in helium radiographs (up to 5-6 mm of WET) due to better secondary filtering. The proton and helium CT images reflected this, with similar noise at the center of the phantom (0.02 RSP) for the proton images and an RSP noise of 0.03 for the proposed filter and 0.06 for the 3σ filter in the helium images. Images reconstructed from data with a dose reduction, up to a factor of 9, maintained a lower noise level using the Prior Filter over the state-of-the-art filtering method. CONCLUSIONS: The proposed filter results in images with equal or reduced noise compared to those that have undergone a filtering method typical of current particle imaging studies. This work also demonstrates that the proposed filter maintains better performance against the state of the art with up to a nine-fold dose reduction.

A novel proton-integrating radiography system design using a monolithic scintillator detector: experimental studies.

Research on proton-based imaging systems aims to improve treatment planning, internal anatomy visualization, and patient alignment for proton radiotherapy. The purpose of this study was to demonstrate a new proton radiography system design consisting of a monolithic plastic scintillator volume and two optical cameras for use with scanning proton pencil beams. Unlike the thin scintillating plates currently used for proton radiography, the plastic scintillator volume (20 × 20 × 20 cm3) captures a wider distribution of proton beam energy depositions and avoids proton-beam modulation. The proton imaging system's characteristics were tested using image uniformity (2.6% over a 5 × 5 cm2 area), stability (0.37%), and linearity (R2 = 1) studies. We used the light distribution produced within the plastic scintillator to generate proton radiographs via two different approaches: (a) integrating light by using a camera placed along the beam axis, and (b) capturing changes to the proton Bragg peak positions with a camera placed perpendicularly to the beam axis. The latter method was used to plot and evaluate relative shifts in percentage depth light (PDL) profiles of proton beams with and without a phantom in the beam path. A curvelet minimization algorithm used differences in PDL profiles to reconstruct and refine the phantom water-equivalent thickness (WET) map. Gammex phantoms were used to compare the proton radiographs generated by these two methods. The relative accuracies in calculating WET of the phantoms using the calibration-based beam-integration (and the PDL) methods were -0.18 ± 0.35% (-0.29 ± 3.11%), -0.11 ± 0.51% (-0.15 ± 2.64%), -2.94 ± 1.20% (-0.75 ± 6.11%), and -1.65 ± 0.35% (0.36 ± 3.93%) for solid water, adipose, cortical bone, and PMMA, respectively. Further exploration of this unique multicamera-based imaging system is warranted and could lead to clinical applications that improve treatment planning and patient alignment for proton radiotherapy.

Experimental comparison of photon versus particle computed tomography to predict tissue relative stopping powers.

PURPOSE: Measurements comparing relative stopping power (RSP) accuracy of state-of-the-art systems representing single-energy and dual-energy computed tomography (SECT/DECT) with proton CT (pCT) and helium CT (HeCT) in biological tissue samples. METHODS: We used 16 porcine and bovine samples of various tissue types and water, covering an RSP range from 0.90 ± 0.06 to 1.78 ± 0.05. Samples were packed and sealed into 3D-printed cylinders ( d = 2  cm, h = 5  cm) and inserted into an in-house designed cylindrical polymethyl methacrylate (PMMA) phantom ( d = 10  cm, h = 10  cm). We scanned the phantom in a commercial SECT and DECT (120 kV; 100  and 140 kV/Sn (tin-filtered)); and acquired pCT and HeCT ( E ∼ 200  MeV/u, 2 ∘ steps, ∼ 6.2 × 10 6 (p)/ ∼ 2.3 × 10 6 (He) particles/projection) with a particle imaging prototype. RSP maps were calculated from SECT/DECT using stoichiometric methods and from pCT/HeCT using the DROP-TVS algorithm. We estimated the average RSP of each tissue per modality in cylindrical volumes of interest and compared it to ground truth RSP taken from peak-detection measurements. RESULTS: Throughout all samples, we observe the following root-mean-squared RSP prediction errors ± combined uncertainty from reference measurement and imaging: SECT 3.10 ± 2.88%, DECT 0.75 ± 2.80%, pCT 1.19 ± 2.81%, and HeCT 0.78 ± 2.81%. The largest mean errors ± combined uncertainty per modality are SECT 8.22 ± 2.79% in cortical bone, DECT 1.74 ± 2.00% in back fat, pCT 1.80 ± 4.27% in bone marrow, and HeCT 1.37 ± 4.25% in bone marrow. Ring artifacts were observed in both pCT and HeCT reconstructions, imposing a systematic shift to predicted RSPs. CONCLUSION: Comparing state-of-the-art SECT/DECT technology and a pCT/HeCT prototype, DECT provided the most accurate RSP prediction, closely followed by particle imaging. The novel modalities pCT and HeCT have the potential to further improve on RSP accuracies with work focusing on the origin and correction of ring artifacts. Future work will study accuracy of proton treatment plans using RSP maps from investigated imaging modalities.

Assessment of the impact of CT calibration procedures for proton therapy planning on pediatric treatments.

PURPOSE: Relative stopping powers (RSPs) for proton therapy are estimated using single-energy computed tomography (SECT), calibrated with standardized tissues of the adult male. It is assumed that those tissues are representative of tissues of all age and sex. Female, male, and pediatric tissues differ from one another in density and composition. In this study, we use tabulated pediatric tissues and computational phantoms to investigate the impact of this assumption on pediatric proton therapy. The potential of dual-energy CT (DECT) to improve the accuracy of these calculations is explored. METHODS: We study 51 human body tissues, categorized into male/female for the age groups newborn, 1-, 5-, 10-, and 15-year-old children, and adult, with given compositions and densities. CT numbers are simulated and RSPs are estimated using SECT and DECT methods. Estimated tissue RSPs from each method are compared to theoretical RSPs. The dose and range errors of each approach are evaluated on three computational phantoms (Ewing's sarcoma, salivary sarcoma, and glioma) derived from pediatric proton therapy patients. RESULTS: With SECT, soft tissues have mean estimation errors and standard deviation up to (1.96 ± 4.18)% observed in newborns, compared to (0.20 ± 1.15)% in adult males. Mean estimation errors for bones are up to (-3.35 ± 4.76)% in pediatrics as opposed to (0.10 ± 0.66)% in adult males. With DECT, mean errors reduce to (0.17 ± 0.13)% and (0.23 ± 0.22)% in newborns (soft tissues/bones). With SECT, dose errors in a Ewing's sarcoma phantom are exceeding 5 Gy (10% of prescribed dose) at the distal end of the treatment field, with volumes of dose errors >5 Gy of V diff > 5 = 4630.7  mm3 . Similar observations are made in the head and neck phantoms, with overdoses to healthy tissue exceeding 2 Gy (4%). A systematic Bragg peak shift resulting in either over- or underdosage of healthy tissues and target volumes depending on the crossed tissues RSP prediction errors is observed. Water equivalent range errors of single beams are between -1.53 and 5.50 mm (min, max) (Ewing's sarcoma phantom), -0.78 and 3.62 mm (salivary sarcoma phantom), and -0.43 and 1.41 mm (glioma phantom). DECT can reduce dose errors to <1 Gy and range errors to <1 mm. CONCLUSION: Single-energy computed tomography estimates RSPs for pediatric tissues with systematic shifts. DECT improves the accuracy of RSPs and dose distributions in pediatric tissues compared to the SECT calibration curve based on adult male tissues.

Image quality evaluation of projection- and depth dose-based approaches to integrating proton radiography using a monolithic scintillator detector.

The purpose of this study is to compare the image quality of an integrating proton radiography (PR) system, composed of a monolithic scintillator and two digital cameras, using integral lateral-dose and integral depth-dose image reconstruction techniques. Monte Carlo simulations were used to obtain the energy deposition in a 3D monolithic scintillator detector (30 × 30 × 30 cm3poly vinyl toluene organic scintillator) to create radiographs of various phantoms-a slanted aluminum cube for spatial resolution analysis and a Las Vegas phantom for contrast analysis. The light emission of the scintillator was corrected using Birks scintillation model. We compared two integrating PR methods and the expected results from an idealized proton tracking radiography system. Four different image reconstruction methods were utilized in this study: integral scintillation light projected from the beams-eye view, depth-dose based reconstruction methods both with and without optimization, and single particle tracking PR was used for reference data. Results showed that heterogeneity artifact due to medium-interface mismatch was identified from the Las Vegas phantom simulated in air. Spatial resolution was found to be highest for single-event reconstruction. Contrast levels, ranked from best to worst, were found to correspond to particle tracking, optimized depth-dose, depth-dose, and projection-based image reconstructions. The image quality of a monolithic scintillator integrating PR system was sufficient to warrant further exploration. These results show promise for potential clinical use as radiographic techniques for visualizing internal patient anatomy during proton radiotherapy.

Machine learning for proton path tracking in proton computed tomography.

A Machine Learning approach to the problem of calculating the proton paths inside a scanned object in proton Computed Tomography is presented. The method is developed in order to mitigate the loss in both spatial resolution and quantitative integrity of the reconstructed images caused by multiple Coulomb scattering of protons traversing the matter. Two Machine Learning models were used: a forward neural network (NN) and the XGBoost method. A heuristic approach, based on track averaging was also implemented in order to evaluate the accuracy limits on track calculation, imposed by the statistical nature of the scattering. Synthetic data from anthropomorphic voxelized phantoms, generated by the Monte Carlo (MC) Geant4 code, were utilized to train the models and evaluate their accuracy, in comparison to a widely used analytical method that is based on likelihood maximization and Fermi-Eyges scattering model. Both NN and XGBoost model were found to perform very close or at the accuracy limit, further improving the accuracy of the analytical method (by 12% in the typical case of 200 MeV protons on 20 cm of water object), especially for protons scattered at large angles. Inclusion of the material information along the path in terms of radiation length did not show improvement in accuracy for the phantoms simulated in the study. A NN was also constructed to predict the error in path calculation, thus enabling a criterion to filter out proton events that may have a negative effect on the quality of the reconstructed image. By parametrizing a large set of synthetic data, the Machine Learning models were proved capable to bring-in an indirect and time efficient way-the accuracy of the MC method into the problem of proton tracking.

Statistical limitations in ion imaging.

In this study, we investigated the capacity of various ion beams available for radiotherapy to produce high quality relative stopping power map acquired from energy-loss measurements. The image quality metrics chosen to compare the different ions were signal-to-noise ratio (SNR) as a function of dose and spatial resolution. Geant4 Monte Carlo simulations were performed for: hydrogen, helium, lithium, boron and carbon ion beams crossing a 20 cm diameter water phantom to determine SNR and spatial resolution. It has been found that protons possess a significantly larger SNR when compared with other ions at a fixed range (up to 36% higher than helium) due to the proton nuclear stability and low dose per primary. However, it also yields the lowest spatial resolution against all other ions, with a resolution lowered by a factor 4 compared to that of carbon imaging, for a beam with the same initial range. When comparing for a fixed spatial resolution of 10 lp cm-1, carbon ions produce the highest image quality metrics with proton ions producing the lowest. In conclusion, it has been found that no ion can maximize all image quality metrics simultaneously and that a choice must be made between spatial resolution, SNR, and dose.

Molière maximum likelihood proton path estimation approximated by cubic Bézier curve for scatter corrected proton CT reconstruction.

A maximum likelihood approach to the problem of calculating the proton paths inside the scanned object in proton computed tomography is presented. Molière theory is used for the first time to derive a physical model that describes proton multiple Coulomb scattering, avoiding the need for the Gaussian approximation currently used. To enable this, the proposed method approximates proton paths with cubic Bézier curves and subsequently maximizes the path likelihood through parametric optimization, based on the Molière model. Results from the Highland formula-based Gaussian approximation are also presented for comparison. The simplex method is utilized for optimisation. The scattering properties of the material(s) of the scanned object are taken into account by appropriately calculating the scattering parameters from the stopping power map that is calculated/updated at every iteration of the algebraic reconstruction process. Proton track length constraint imposed by the proton energy loss is accounted for. The method is also applied in the case that no exit angle data are measured. Geant4 Monte Carlo simulations were performed for model validation. Our results show that use of Molière probability density function for modelling the multiple Coulomb scattering presents a modest 2% accuracy improvement over the Gaussian approximation and most-likely-path method. Simulations of voxelized phantom showed no essential benefit from the inclusion of the material information into the optimization, while path optimization with energy constraint slightly increased path resolution in a bone/water interface phantom. Method error was found to depend on energy, proton track-length within the medium, and proportion of data filtering.

A comparison of direct reconstruction algorithms in proton computed tomography.

Several direct algorithms have been proposed to take into account the non-linear path of protons in the reconstruction of a proton CT (pCT) image. This paper presents a comparison between five of them, in terms of spatial resolution and relative stopping power (RSP) accuracy. Our comparison includes (1) a distance-driven algorithm extending the filtered backprojection to non-linear trajectories (DD), (2) an algorithm reconstructing a pCT image from optimized projections (ML), (3) a backproject-then-filter approach using a 2D cone filter (BTF), (4) a differentiated backprojection algorithm based on the inversion of the Hilbert transform (DBP), and (5) an algorithm using a 2D directional ramp filter (DR). We have simulated a single tracking pCT set-up using Geant4 through GATE, with a proton source and two position, direction and energy detectors upstream and downstream from the object. Tracker uncertainties were added on the position and direction measurements. A Catphan 528 phantom and a spiral phantom were simulated to measure the spatial resolution and a Gammex 467 phantom was used for the RSP accuracy. Each proton's trajectory was estimated using a most likely path (MLP) formalism. The spatial resolution was evaluated using the frequency corresponding to a modulation transfer function of 10% of its peak value and the RSP accuracy using the mean values in the inserts of the Gammex phantom. In terms of spatial resolution, it was shown that, for ideal trackers, the DR and BTF methods offer a slightly better resolution since each proton is directly binned in the image grid according to its MLP. However, all methods but the ML show comparable resolution when using realistic trackers. Regarding the RSP, three algorithms (DR, DD and BTF) show a mean relative error inside the inserts about 0.1%. As the DR and BTF methods are more computationally expensive, the DD-which allows the same spatial resolution in realistic conditions and the same accuracy-and the DBP-which has a fairly good accuracy (<0.2%) and allows reconstruction from truncated data-can be used for a reduced reconstruction time.

Statistical limitations in proton imaging.

Proton imaging is a promising technology for proton radiotherapy as it can be used for: (1) direct sampling of the tissue stopping power, (2) input information for multi-modality RSP reconstruction, (3) gold-standard calibration against concurrent techniques, (4) tracking motion and (5) pre-treatment positioning. However, no end-to-end characterization of the image quality (signal-to-noise ratio and spatial resolution, blurring uncertainty) against the dose has been done. This work aims to establish a model relating these characteristics and to describe their relationship with proton energy and object size. The imaging noise originates from two processes: the Coulomb scattering with the nucleus, producing a path deviation, and the energy loss straggling with electrons. The noise is found to increases with thickness crossed and, independently, decreases with decreasing energy. The scattering noise is dominant around high-gradient edge whereas the straggling noise is maximal in homogeneous regions. Image quality metrics are found to behave oppositely against energy: lower energy minimizes both the noise and the spatial resolution, with the optimal energy choice depending on the application and location in the imaged object. In conclusion, the model presented will help define an optimal usage of proton imaging to reach the promised application of this technology and establish a fair comparison with other imaging techniques.

Theoretical considerations on the spatial resolution limit of single-event particle radiography.

Particle radiography (pRad) has been proposed and investigated as a promising tool for particle therapy as it provides a water equivalent thickness (WET) image of the patient. In single-event particle imaging, for each measured particle, the related most likely path (MLP) through the object is estimated to account for multiple Coulomb scattering (MCS). In previous studies, the accuracy limit of the MLP has been used to determine the spatial resolution limit. In this work, we investigate the limit of the spatial resolution achievable with different pRad algorithms based on a theoretical model of the particle scattering for an ideal beam and detector. First, we investigate binning the particles in a plane seated at the depth of the object of interest (plane-of-interest binning; PIB) and extend existing theoretical considerations also to objects not located in the binning plane. We use this to model the spatial resolution in case of binning the particles directly at the front or rear tracker (FTB and RTB, respectively). Further, we investigate evenly distributing the particles' WET along their trajectory into pixel channels and creating the pRad image as channel mean (along-path-binning; APB). Monte Carlo simulations are used to qualitatively investigate the different algorithms and to validate the theoretical predictions. We show that projecting the scattered particle paths onto a single image will inevitably result in a limited spatial resolution lower than expected from only the MLP uncertainty. Only in the case where the depth of a feature is known and used as binning depth for PIB, the spatial resolution of that feature is equal to the path estimation accuracy. For the APB algorithm the spatial resolution decreases with increasing depth in the object, especially if the true particle path through the object would be known. The derived theoretical models will be useful for future development of improved pRad reconstruction algorithms.

The impact of secondary fragments on the image quality of helium ion imaging.

Single-event ion imaging enables the direct reconstruction of the relative stopping power (RSP) information required for ion-beam therapy. Helium ions were recently hypothesized to be the optimal species for such technique. The purpose of this work is to investigate the effect of secondary fragments on the image quality of helium CT (HeCT) and to assess the performance of a prototype proton CT (pCT) scanner when operated with helium beams in Monte Carlo simulations and experiment. Experiments were conducted installing the U.S. pCT consortium prototype scanner at the Heidelberg Ion-Beam Therapy Center (HIT). Simulations were performed with the scanner using the TOPAS toolkit. HeCT images were reconstructed for a cylindrical water phantom, the CTP404 (sensitometry), and the CTP528 (line-pair) [Formula: see text] ® modules. To identify and remove individual events caused by fragmentation, the multistage energy detector of the scanner was adapted to function as a [Formula: see text] telescope. The use of the developed filter eliminated the otherwise arising ring artifacts in the HeCT reconstructed images. For the HeCT reconstructed images of a water phantom, the maximum RSP error was improved by almost a factor 8 with respect to unfiltered images in the simulation and a factor 10 in the experiment. Similarly, for the CTP404 module, the mean RSP accuracy improved by a factor 6 in both the simulation and the experiment when the filter was applied (mean relative error 0.40% in simulation, 0.45% in experiment). In the evaluation of the spatial resolution through the CTP528 module, the main effect of the filter was noise reduction. For both simulated and experimental images the spatial resolution was ∼4 lp cm-1. In conclusion, the novel filter developed for secondary fragments proved to be effective in improving the visual quality and RSP accuracy of the reconstructed images. With the filter, the pCT scanner is capable of accurate HeCT imaging.

Extension of the Fermi-Eyges most-likely path in heterogeneous medium with prior knowledge information.

Particle imaging suffers from poor spatial resolution due to the multiple Coulomb scattering deflections undergone by the particles throughout their path. To account for these deflections, a most-likely path (MLP) formalism was developed based on a Bayesian adaption of the Fermi-Eyges theory. Previous work calculated the MLP formalism in a homogeneous water medium as an initial estimate. However, this potentially reduces the accuracy of the MLP estimate as well as the achievable resolution of the subsequent tomographic reconstruction. This work investigates the potential gain of introducing prior-knowledge on the medium composition and density to improve the MLP accuracy. To do so, a Monte Carlo (MC) Geant4 algorithm was used to simulate protons ([Formula: see text]) crossing three different anthropomorphic phantoms representing the lung, abdomen, and head. The prior-knowledge information is gathered from (1) the MC simulation for ground-truth (MLP-GT), or from (2) a recent DECT material decomposition technique (MLP-DECT). The reconstructed path accuracy using prior-knowledge methods is compared with (3) the path reconstructed in homogeneous water (MLP-Water) and (4) a path reconstruction method where the proton path is projected onto a Hull at the boundary of the phantom with a subsequent MLP-Water calculation (MLP-Hull). For each path reconstruction method, the maximal root-mean-square error (RMSmax) is compared between the reconstructed and the MC path. In every phantom, the RMSmax is decreased between the MLP-Water and the three other path algorithms that take into account heterogeneities ([Formula: see text] for the lung, [Formula: see text] for the abdomen and [Formula: see text] for the head), with no significant differences between each (MLP-DECT, MLP-GT and MLP-Hull). In conclusion, the introduction of prior-knowledge in the MLP formalism decreases the RMS uncertainty to the MC path, but no further than the use of a simpler Hull contour algorithm. The use of this Hull algorithm is suggested for future particle imaging applications.

Pre-treatment patient-specific stopping power by combining list-mode proton radiography and x-ray CT.

The relative stopping power (RSP) uncertainty is the largest contributor to the range uncertainty in proton therapy. The purpose of this work was to develop a systematic method that yields accurate and patient-specific RSPs by combining (1) pre-treatment x-ray CT and (2) daily proton radiography of the patient. The method was formulated as a penalized least squares optimization problem (argmin([Formula: see text])). The parameter A represents the cumulative path-length crossed by the proton in each material, separated by thresholding on the HU. The material RSPs (water equivalent thickness/physical thickness) are denoted by x. The parameter b is the list-mode proton radiography produced using Geant4 simulations. The problem was solved using a non-negative linear-solver with [Formula: see text]. A was computed by superposing proton trajectories calculated with a cubic or linear spline approach to the CT. The material's RSP assigned in Geant4 were used for reference while the clinical HU-RSP calibration curve was used for comparison. The Gammex RMI-467 phantom was first investigated. The standard deviation between the estimated material RSP and the calculated RSP is 0.45%. The robustness of the techniques was then assessed as a function of the number of projections and initial proton energy. Optimization with two initial projections yields precise RSP (⩽1.0%) for 330 MeV protons. 250 MeV protons have shown higher uncertainty (⩽2.0%) due to the loss of precision in the path estimate. Anthropomorphic phantoms of the head, pelvis, and lung were subsequently evaluated. Accurate RSP has been obtained for the head ([Formula: see text]), the lung ([Formula: see text]) and the pelvis ([Formula: see text]). The range precision has been optimized using the calibration curves obtained with the algorithm, yielding a mean [Formula: see text] difference to the reference of 0.11 ±0.09%, 0.28 ± 0.34% and [Formula: see text] in the same order. The solution's accuracy is limited by the assumed HU/RSP bijection, neglecting inherent degeneracy. The proposed formulation of the problem with prior knowledge x-ray CT demonstrates potential to increase the accuracy of present RSP estimates.

A theoretical framework to predict the most likely ion path in particle imaging.

In this work, a generic rigorous Bayesian formalism is introduced to predict the most likely path of any ion crossing a medium between two detection points. The path is predicted based on a combination of the particle scattering in the material and measurements of its initial and final position, direction and energy. The path estimate's precision is compared to the Monte Carlo simulated path. Every ion from hydrogen to carbon is simulated in two scenarios, (1) where the range is fixed and (2) where the initial velocity is fixed. In the scenario where the range is kept constant, the maximal root-mean-square error between the estimated path and the Monte Carlo path drops significantly between the proton path estimate (0.50 mm) and the helium path estimate (0.18 mm), but less so up to the carbon path estimate (0.09 mm). However, this scenario is identified as the configuration that maximizes the dose while minimizing the path resolution. In the scenario where the initial velocity is fixed, the maximal root-mean-square error between the estimated path and the Monte Carlo path drops significantly between the proton path estimate (0.29 mm) and the helium path estimate (0.09 mm) but increases for heavier ions up to carbon (0.12 mm). As a result, helium is found to be the particle with the most accurate path estimate for the lowest dose, potentially leading to tomographic images of higher spatial resolution.

Stopping power accuracy and achievable spatial resolution of helium ion imaging using a prototype particle CT detector system.

A precise relative stopping power map of the patient is crucial for accurate particle therapy. Charged particle imaging determines the stopping power either tomographically - particle computed tomography (pCT) - or by combining prior knowledge from particle radiography (pRad) and x-ray CT. Generally, multiple Coulomb scattering limits the spatial resolution. Compared to protons, heavier particles scatter less due to their lower charge/mass ratio. A theoretical framework to predict the most likely trajectory of particles in matter was developed for light ions up to carbon and was found to be the most accurate for helium comparing for fixed initial velocity. To further investigate the potential of helium in particle imaging, helium computed tomography (HeCT) and radiography (HeRad) were studied at the Heidelberg Ion-Beam Therapy Centre (HIT) using a prototype pCT detector system registering individual particles, originally developed by the U.S. pCT collaboration. Several phantoms were investigated: modules of the Catphan QA phantom for analysis of spatial resolution and achievable stopping power accuracy, a paediatric head phantom (CIRS) and a custommade phantom comprised of animal meat enclosed in a 2 % agarose mixture representing human tissue. The pCT images were reconstructed applying the CARP iterative reconstruction algorithm. The MTF10% was investigated using a sharp edge gradient technique. HeRad provides a spatial resolution above that of protons (MTF1010%=6.07 lp/cm for HeRad versus MTF10%=3.35 lp/cm for proton radiography). For HeCT, the spatial resolution was limited by the number of projections acquired (90 projections for a full scan). The RSP accuracy for all inserts of the Catphan CTP404 module was found to be 2.5% or better and is subject to further optimisation. In conclusion, helium imaging appears to offer higher spatial resolution compared to proton imaging. In future studies, the advantage of helium imaging compared to other imaging modalities in clinical applications will be further explored.

A maximum likelihood method for high resolution proton radiography/proton CT.

Multiple Coulomb scattering (MCS) is the largest contributor to blurring in proton imaging. In this work, we developed a maximum likelihood least squares estimator that improves proton radiography's spatial resolution. The water equivalent thickness (WET) through projections defined from the source to the detector pixels were estimated such that they maximizes the likelihood of the energy loss of every proton crossing the volume. The length spent in each projection was calculated through the optimized cubic spline path estimate. The proton radiographies were produced using Geant4 simulations. Three phantoms were studied here: a slanted cube in a tank of water to measure 2D spatial resolution, a voxelized head phantom for clinical performance evaluation as well as a parametric Catphan phantom (CTP528) for 3D spatial resolution. Two proton beam configurations were used: a parallel and a conical beam. Proton beams of 200 and 330 MeV were simulated to acquire the radiography. Spatial resolution is increased from 2.44 lp cm-1 to 4.53 lp cm-1 in the 200 MeV beam and from 3.49 lp cm-1 to 5.76 lp cm-1 in the 330 MeV beam. Beam configurations do not affect the reconstructed spatial resolution as investigated between a radiography acquired with the parallel (3.49 lp cm-1 to 5.76 lp cm-1) or conical beam (from 3.49 lp cm-1 to 5.56 lp cm-1). The improved images were then used as input in a photon tomography algorithm. The proton CT reconstruction of the Catphan phantom shows high spatial resolution (from 2.79 to 5.55 lp cm-1 for the parallel beam and from 3.03 to 5.15 lp cm-1 for the conical beam) and the reconstruction of the head phantom, although qualitative, shows high contrast in the gradient region. The proposed formulation of the optimization demonstrates serious potential to increase the spatial resolution (up by 65[Formula: see text]) in proton radiography and greatly accelerate proton computed tomography reconstruction.