Rapid calculation of detectability in Bayesian single photon emission computed tomography

We consider the calculation of lesion detectability using a mathematical model observer, the channelized Hotelling observer (CHO), in a signal-known-exactly/background-known-exactly detection task for single photon emission computed tomography (SPECT). We focus on SPECT images reconstructed with Bayesian maximum a posteriori methods. While model observers are designed to replace time-consuming studies using human observers, the calculation of CHO detectability is usually accomplished using a large number of sample images, which is still time consuming. We develop theoretical expressions for a measure of detectability, the signal-to-noise-ratio (SNR) of a CHO observer, that can be very rapidly evaluated. Key to our expressions are approximations to the reconstructed image covariance. In these approximations, we use methods developed in the PET literature, but modify them to reflect the different nature of attenuation and distance-dependent blur in SPECT. We validate our expressions with Monte Carlo methods. We show that reasonably accurate estimates of the SNR can be obtained at a computational expense equivalent to approximately two projection operations, and that evaluating SNR for subsequent lesion locations requires negligible additional computation.     

Determination of radiotherapy X-ray spectra using a screen-film system

A method to determine the X-ray spectrum delivered by a medical linear accelerator is presented. This method consists of an analytical calculation of the primary spectrum using the Schiff bremsstrahlung cross-section formula. A correction factor that accounts for the scatter component of the spectrum is estimated by comparing the signal in two screen-film systems to a theoretical prediction using a model of energy deposition in such detectors. The model makes use of the quantum absorption efficiency and the average energy deposited per interacting photon concepts. These two quantities are calculated by means of Monte Carlo simulations of the screen-film systems used. This method is capable of determining the spectrum as a function of the spatial position across a plane perpendicular to the beam central axis. It does not, however, render information about the direction cosines of the X-ray fluence crossing such a plane, a requirement in order to produce a full phase-space file that can be used in conjunction with a Monte Carlo dose calculation engine.     

Partial signal space projection for artefact removal in MEG measurements: a theoretical analysis.

Standard methods for artefact removal in MEG or EEG measurements consist of rejection of either corrupted epochs or signal space projection (SSP). We propose to combine the two methods by applying SSP only in corrupted epochs and thus using both temporal and spatial information. This partial signal space projection necessarily results in smaller variances for the source localization. Formulae for dipole localization errors as a function of fraction of corrupted epochs are derived and verified in Monte Carlo simulations of MEG measurements corrupted with eye artefacts. A theoretical analysis of various measuring devices, classes of artefact and locations of dipole of interest shows that the proposed method leads to significant improvement for frontal signal dipoles and for 30-80% corrupted epochs.     

Improved scatter correction using adaptive scatter kernel superposition

Accurate scatter correction is required to produce high-quality reconstructions of x-ray cone-beam computed tomography (CBCT) scans. This paper describes new scatter kernel superposition (SKS) algorithms for deconvolving scatter from projection data. The algorithms are designed to improve upon the conventional approach whose accuracy is limited by the use of symmetric kernels that characterize the scatter properties of uniform slabs. To model scatter transport in more realistic objects, nonstationary kernels, whose shapes adapt to local thickness variations in the projection data, are proposed. Two methods are introduced: (1) adaptive scatter kernel superposition (ASKS) requiring spatial domain convolutions and (2) fast adaptive scatter kernel superposition (fASKS) where, through a linearity approximation, convolution is efficiently performed in Fourier space. The conventional SKS algorithm, ASKS, and fASKS, were tested with Monte Carlo simulations and with phantom data acquired on a table-top CBCT system matching the Varian On-Board Imager (OBI). All three models accounted for scatter point-spread broadening due to object thickening, object edge effects, detector scatter properties and an anti-scatter grid. Hounsfield unit (HU) errors in reconstructions of a large pelvis phantom with a measured maximum scatter-to-primary ratio over 200% were reduced from -90 ± 58 HU (mean ± standard deviation) with no scatter correction to 53 ± 82 HU with SKS, to 19 ± 25 HU with fASKS and to 13 ± 21 HU with ASKS. HU accuracies and measured contrast were similarly improved in reconstructions of a body-sized elliptical Catphan phantom. The results show that the adaptive SKS methods offer significant advantages over the conventional scatter deconvolution technique.     

Direct aperture optimization for IMRT using Monte Carlo generated beamlets

This work introduces an EGSnrc-based Monte Carlo (MC) beamlet does distribution matrix into a direct aperture optimization (DAO) algorithm for IMRT inverse planning. The technique is referred to as Monte Carlo-direct aperture optimization (MC-DAO). The goal is to assess if the combination of accurate Monte Carlo tissue inhomogeneity modeling and DAO inverse planning will improve the dose accuracy and treatment efficiency for treatment planning. Several authors have shown that the presence of small fields and/or inhomogeneous materials in IMRT treatment fields can cause dose calculation errors for algorithms that are unable to accurately model electronic disequilibrium. This issue may also affect the IMRT optimization process because the dose calculation algorithm may not properly model difficult geometries such as targets close to low-density regions (lung, air etc.). A clinical linear accelerator head is simulated using BEAMnrc (NRC, Canada). A novel in-house algorithm subdivides the resulting phase space into 2.5 X 5.0 mm2 beamlets. Each beamlet is projected onto a patient-specific phantom. The beamlet dose contribution to each voxel in a structure-of-interest is calculated using DOSXYZnrc. The multileaf collimator (MLC) leaf positions are linked to the location of the beamlet does distributions. The MLC shapes are optimized using direct aperture optimization (DAO). A final Monte Carlo calculation with MLC modeling is used to compute the final dose distribution. Monte Carlo simulation can generate accurate beamlet dose distributions for traditionally difficult-to-calculate geometries, particularly for small fields crossing regions of tissue inhomogeneity. The introduction of DAO results in an additional improvement by increasing the treatment delivery efficiency. For the examples presented in this paper the reduction in the total number of monitor units to deliver is approximately 33% compared to fluence-based optimization methods.     

Clinical implementation of full Monte Carlo dose calculation in proton beam therapy

The goal of this work was to facilitate the clinical use of Monte Carlo proton dose calculation to support routine treatment planning and delivery. The Monte Carlo code Geant4 was used to simulate the treatment head setup, including a time-dependent simulation of modulator wheels (for broad beam modulation) and magnetic field settings (for beam scanning). Any patient-field-specific setup can be modeled according to the treatment control system of the facility. The code was benchmarked against phantom measurements. Using a simulation of the ionization chamber reading in the treatment head allows the Monte Carlo dose to be specified in absolute units (Gy per ionization chamber reading). Next, the capability of reading CT data information was implemented into the Monte Carlo code to model patient anatomy. To allow time-efficient dose calculation, the standard Geant4 tracking algorithm was modified. Finally, a software link of the Monte Carlo dose engine to the patient database and the commercial planning system was established to allow data exchange, thus completing the implementation of the proton Monte Carlo dose calculation engine ('DoC++'). Monte Carlo re-calculated plans are a valuable tool to revisit decisions in the planning process. Identification of clinically significant differences between Monte Carlo and pencil-beam-based dose calculations may also drive improvements of current pencil-beam methods. As an example, four patients (29 fields in total) with tumors in the head and neck regions were analyzed. Differences between the pencil-beam algorithm and Monte Carlo were identified in particular near the end of range, both due to dose degradation and overall differences in range prediction due to bony anatomy in the beam path. Further, the Monte Carlo reports dose-to-tissue as compared to dose-to-water by the planning system. Our implementation is tailored to a specific Monte Carlo code and the treatment planning system XiO (Computerized Medical Systems Inc.). However, this work describes the general challenges and considerations when implementing proton Monte Carlo dose calculation in a clinical environment. The presented solutions can be easily adopted for other planning systems or other Monte Carlo codes.     

X-ray imaging optimization using virtual phantoms and computerized observer modelling

This study develops and demonstrates a realistic x-ray imaging simulator with computerized observers to maximize lesion detectability and minimize patient exposure. A software package, ViPRIS, incorporating two computational patient phantoms, has been developed for simulating x-ray radiographic images. A tomographic phantom, VIP-Man, constructed from Visible Human anatomical colour images is used to simulate the scattered portion using the ESGnrc Monte Carlo code. The primary portion of an x-ray image is simulated using the projection ray-tracing method through the Visible Human CT data set. To produce a realistic image, the software simulates quantum noise, blurring effects, lesions, detector absorption efficiency and other imaging artefacts. The primary and scattered portions of an x-ray chest image are combined to form a final image for computerized observer studies and image quality analysis. Absorbed doses in organs and tissues of the segmented VIP-Man phantom were also obtained from the Monte Carlo simulations. Approximately 25,000 simulated images and 2,500,000 data files were analysed using computerized observers. Hotelling and Laguerre-Gauss Hotelling observers are used to perform various lesion detection tasks. Several model observer tasks were used including SKE/BKE, MAFC and SKEV. The energy levels and fluence at the minimum dose required to detect a small lesion were determined with respect to lesion size, location and system parameters.     

Modelling of electron contamination in clinical photon beams for Monte Carlo dose calculation

The purpose of this work is to model electron contamination in clinical photon beams and to commission the source model using measured data for Monte Carlo treatment planning. In this work, a planar source is used to represent the contaminant electrons at a plane above the upper jaws. The source size depends on the dimensions of the field size at the isocentre. The energy spectra of the contaminant electrons are predetermined using Monte Carlo simulations for photon beams from different clinical accelerators. A 'random creep' method is employed to derive the weight of the electron contamination source by matching Monte Carlo calculated monoenergetic photon and electron percent depth-dose (PDD) curves with measured PDD curves. We have integrated this electron contamination source into a previously developed multiple source model and validated the model for photon beams from Siemens PRIMUS accelerators. The EGS4 based Monte Carlo user code BEAM and MCSIM were used for linac head sinulation and dose calculation. The Monte Carlo calculated dose distributions were compared with measured data. Our results showed good agreement (less than 2% or 2 mm) for 6, 10 and 18 MV photon beams.     

Simulation study of second-harmonic microscopic imaging signals through tissue-like turbid media

We establish, for the first time, a simulation model for dealing with the second-harmonic signals under a microscope through a tissue-like turbid medium, based on the Monte Carlo method. With this model, the angle-resolved distribution and the signal level eta of second-harmonic light through a slab of the turbid medium are demonstrated and the effects of the thickness (d) of the turbid medium, the numerical aperture (NA) of the objective as well as the size (rho) of the scatterers forming the turbid medium are explored. Simulation results reveal that the use of a small objective NA results in a narrow angle distribution but strong second-harmonic signals. A turbid medium consisting of large scattering particles has a strong influence on the angle distribution and the signal level eta, which results in a low penetration limit for second-harmonic signals made up of ballistic photons. It is approximately 30 microm in our situation.     

Combining deterministic and Monte Carlo calculations for fast estimation of scatter intensities in CT

A side effect of increased volume coverage by using multi-row and flat-panel detectors in computed tomography (CT) is the concurrently growing contribution of scattered radiation to the measured signal. In order to investigate the effect of scatter on x-ray projections used for CT imaging, our study aimed at the development of a simulation tool for fast calculation of primary and scatter intensities. We developed a deterministic method to assess the contribution of single-scatter events to the measured signal. The investigation of multiple scatter by Monte Carlo simulations showed that it results in a smooth signal as compared to single scatter. A hybrid method is proposed in order to optimize the performance of the scatter simulation: a fast and exact analytical calculation of the single-scatter intensity combined with a coarse Monte Carlo (MC) estimate of multiple scatter to reduce overall computational expenses, while assuring an acceptable signal quality. The results of the hybrid simulation of total scatter were in excellent agreement with the corresponding MC only simulations, thereby allowing us to reduce computational time by orders of magnitude. Estimates of two-dimensional scatter distributions for flat-panel CT imaging took about 30-40 s (per projection). The hybrid method provides a realistic simulation of x-ray scatter and offers a basis for scatter correction approaches.     

Experimental determination of electron source parameters for accurate Monte Carlo calculation of large field electron therapy

Extensive work has been performed to validate Monte Carlo models for both photon and electron beams under standard conditions. However, for large field electron beam therapy, Monte Carlo simulations have not been able to provide good agreement when compared to the measured dose distributions. Since the accuracy of the calculation relies heavily on the geometry parameters of the linear accelerator and the characteristics of the incident electron beam, it is crucial to have a complete comprehension of these independent factors. In this work, the electron focal spot size for a CL21EX linac with various energies (6, 9, 12 and 16 MeV) was measured with a slit camera composed of alternating lead and paper sheets. For all the energies investigated, the electron focal spot is found to be elliptical and has a full width at half maximum (FWHM) ranging from 1.69 mm to 2.24 mm. A shift with respect to the crosshair was associated with each measured focal spot. In addition, we present an improved result for the large field in-air profile by utilizing a proposed divergent beam model in conjunction with the experimental focal spot dimension. This model can potentially provide a solution to the Monte Carlo validation of large field electron beam therapy.     

Amending of fluorescence sensor signal localization in human skin by matching of the refractive index

Fluorescence diagnostic techniques are notable amongst many other optical methods because they offer high sensitivity and noninvasive measurement of tissue properties. However, a combination of multiple scattering and physical heterogeneity of biological tissues hampers interpretation of the fluorescence measurements. Analyses of the spatial distribution of endogenous and exogenous fluorophores excitation within tissues and their contribution to the detected signal localization are essential for many applications. We have developed a novel Monte Carlo technique that gives a graphical perception of how the excitation and fluorescence detected signal are localized in tissues. Our model takes into account the spatial distribution of fluorophores, the variation of concentrations and quantum yield. We demonstrate that matching the refractive indices of the ambient medium and topical skin layer improves spatial localization of the detected fluorescence signal within the tissues.     

Four-dimensional Monte Carlo simulation of time-dependent geometries

Dynamic radiation therapy techniques require that the treatment head setup varies in time and space during dose delivery. Examples are leaf motion in intensity modulated x-ray therapy, modulator wheel rotation in conventional proton therapy and variable magnet settings in proton beam scanning. Consequently, for Monte Carlo dose calculation the results of several independent simulations usually have to be combined. Depending on the complexity and the required accuracy this can become cumbersome. We present a technique to simulate time-dependent geometries within a single four-dimensional Monte Carlo simulation using the GEANT4 Monte Carlo package. Results for proton therapy applications are shown.     

Nonparametric ROC and LROC analysis

In this paper we review several results of the nonparametric receiver operating characteristic (ROC) analysis and present an extension to the nonparametric localization ROC (LROC) analysis. Equations for the estimation of the area under the characteristic curve and for the variance calculations are derived. Expressions for the choice of the optimal ratio between the number of signal-absent and signal-present image samples are also presented. The results can be applied both with continuous or discrete scoring scales. The simulation studies carried out validate the theoretical derivations and show that the LROC analysis is considerably more sensitive than the ROC analysis.     

Accelerating Monte Carlo simulations of radiation therapy dose distributions using wavelet threshold de-noising

The Monte Carlo dose calculation method works by simulating individual energetic photons or electrons as they traverse a digital representation of the patient anatomy. However, Monte Carlo results fluctuate until a large number of particles are simulated. We propose wavelet threshold de-noising as a postprocessing step to accelerate convergence of Monte Carlo dose calculations. A sampled rough function (such as Monte Carlo noise) gives wavelet transform coefficients which are more nearly equal in amplitude than those of a sampled smooth function. Wavelet hard-threshold de-noising sets to zero those wavelet coefficients which fall below a threshold; the image is then reconstructed. We implemented the computationally efficient 9,7-biorthogonal filters in the C language. Transform results were averaged over transform origin selections to reduce artifacts. A method for selecting best threshold values is described. The algorithm requires about 336 floating point arithmetic operations per dose grid point. We applied wavelet threshold de-noising to two two-dimensional dose distributions: a dose distribution generated by 10 MeV electrons incident on a water phantom with a step-heterogeneity, and a slice from a lung heterogeneity phantom. Dose distributions were simulated using the Integrated Tiger Series Monte Carlo code. We studied threshold selection, resulting dose image smoothness, and resulting dose image accuracy as a function of the number of source particles. For both phantoms, with a suitable value of the threshold parameter, voxel-to-voxel noise was suppressed with little introduction of bias. The roughness of wavelet de-noised dose distributions (according to a Laplacian metric) was nearly independent of the number of source electrons, though the accuracy of the de-noised dose image improved with increasing numbers of source electrons. We conclude that wavelet shrinkage de-noising is a promising method for effectively accelerating Monte Carlo dose calculations by factors of 2 or more.     

Establishing an Initial Electron Beam Model with Monte Carlo Simulation for a Single 6 MV X-ray Medical Linac Based on Particle Dynamics Characteristics

Objective：In this study, we try to establish an initial electron beam model by combining Monte Carlo simulation method with particle dynamic calculation （TRSV） for the single 6 MV X-ray accelerating waveguide of BJ-6 medical linac. Methods and Materials ： 1. We adapted the treatment head configuration of BJ-6 medical linac made by Beijing Medical Equipment Institute （BMEI） as the radiation system for this study. 2. Use particle dynamics calculation code called TRSV to drive out the initial electron beam parameters of the energy spectrum, the spatial intensity distribution, and the beam incidence angle. 3. Analyze the 6 MV X-ray beam characteristics of PDDc , OARc in a water phantom by using Monte Carlo simulation （ BEAMnrc, DOSXYZnrc） for a preset of the initial electron beam parameters which have been determined by TRSV, do the comparisons of the measured results of PDDm, OARm in a real water phantom, and then use the deviations of calculated and measured results to slightly modify the initial electron beam model back and forth until the deviations meet the error less than 2%. Results：The deviations between the Monte Carlo simulation results of percentage depth doses at PDDc and off-axis ratios OARc and the measured results of PDDm and OARm in a water phantom were within 2%. Conclusion：When doing the Monte Carlo simulation to determine the parameters of an initial electron beam for a particular medical linac like B J-6, modifying some parameters based on the particle dynamics calculation code would give some more reasonable and more acceptable results.     

A Monte Carlo dose calculation tool for radiotherapy treatment planning

A Monte Carlo user code, MCDOSE, has been developed for radiotherapy treatment planning (RTP) dose calculations. MCDOSE is designed as a dose calculation module suitable for adaptation to host RTP systems. MCDOSE can be used for both conventional photon/electron beam calculation and intensity modulated radiotherapy (IMRT) treatment planning. MCDOSE uses a multiple-source model to reconstruct the treatment beam phase space. Based on Monte Carlo simulated or measured beam data acquired during commissioning, source-model parameters are adjusted through an automated procedure. Beam modifiers such as jaws, physical and dynamic wedges, compensators, blocks, electron cut-outs and bolus are simulated by MCDOSE together with a 3D rectilinear patient geometry model built from CT data. Dose distributions calculated using MCDOSE agreed well with those calculated by the EGS4/DOSXYZ code using different beam set-ups and beam modifiers. Heterogeneity correction factors for layered-lung or layered-bone phantoms as calculated by both codes were consistent with measured data to within 1%. The effect of energy cut-offs for particle transport was investigated. Variance reduction techniques were implemented in MCDOSE to achieve a speedup factor of 10-30 compared to DOSXYZ.