Sequential Monte Carlo for Bayesian sequentially designed experiments for discrete data

In this paper we present a sequential Monte Carlo algorithm for Bayesian sequential experimental design applied to generalised non-linear models for discrete data. The approach is computationally convenient in that the information of newly observed data can be incorporated through a simple re-weighting step. We also consider a flexible parametric model for the stimulus-response relationship together with a newly developed hybrid design utility that can produce more robust estimates of the target stimulus in the presence of substantial model and parameter uncertainty. The algorithm is applied to hypothetical clinical trial or bioassay scenarios. In the discussion, potential generalisations of the algorithm are suggested to possibly extend its applicability to a wide variety of scenarios

A sequential Monte Carlo algorithm to incorporate model uncertainty in Bayesian sequential design

Here we present a sequential Monte Carlo (SMC) algorithm that can be used for any one-at-a-time Bayesian sequential design problem in the presence of model uncertainty where discrete data are encountered. Our focus is on adaptive design for model discrimination but the methodology is applicable if one has a different design objective such as parameter estimation or prediction. An SMC algorithm is run in parallel for each model and the algorithm relies on a convenient estimator of the evidence of each model which is essentially a function of importance sampling weights. Other methods for this task such as quadrature, often used in design, suffer from the curse of dimensionality. Approximating posterior model probabilities in this way allows us to use model discrimination utility functions derived from information theory that were previously difficult to compute except for conjugate models. A major benefit of the algorithm is that it requires very little problem specific tuning. We demonstrate the methodology on three applications, including discriminating between models for decline in motor neuron numbers in patients suffering from neurological diseases such as Motor Neuron disease.

A sequential Monte Carlo approach to the sequential design for discriminating between rival continuous data models

Here we present a sequential Monte Carlo approach to Bayesian sequential design for the incorporation of model uncertainty. The methodology is demonstrated through the development and implementation of two model discrimination utilities; mutual information and total separation, but it can also be applied more generally if one has different experimental aims. A sequential Monte Carlo algorithm is run for each rival model (in parallel), and provides a convenient estimate of the marginal likelihood (of each model) given the data, which can be used for model comparison and in the evaluation of utility functions. A major benefit of this approach is that it requires very little problem specific tuning and is also computationally efficient when compared to full Markov chain Monte Carlo approaches. This research is motivated by applications in drug development and chemical engineering.

Marginal reversible jump Markov chain Monte Carlo with application to motor unit number estimation

Motor unit number estimation (MUNE) is a method which aims to provide a quantitative indicator of progression of diseases that lead to loss of motor units, such as motor neurone disease. However the development of a reliable, repeatable and fast real-time MUNE method has proved elusive hitherto. Ridall et al. (2007) implement a reversible jump Markov chain Monte Carlo (RJMCMC) algorithm to produce a posterior distribution for the number of motor units using a Bayesian hierarchical model that takes into account biological information about motor unit activation. However we find that the approach can be unreliable for some datasets since it can suffer from poor cross-dimensional mixing. Here we focus on improved inference by marginalising over latent variables to create the likelihood. In particular we explore how this can improve the RJMCMC mixing and investigate alternative approaches that utilise the likelihood (e.g. DIC (Spiegelhalter et al., 2002)). For this model the marginalisation is over latent variables which, for a larger number of motor units, is an intractable summation over all combinations of a set of latent binary variables whose joint sample space increases exponentially with the number of motor units. We provide a tractable and accurate approximation for this quantity and also investigate simulation approaches incorporated into RJMCMC using results of Andrieu and Roberts (2009).

Correcting radiation survey data to account for increased leakage during intensity modulated radiotherapy treatments

Purpose Intensity modulated radiotherapy (IMRT) treatments require more beam-on time and produce more linac head leakage to deliver similar doses to conventional, unmodulated, radiotherapy treatments. It is necessary to take this increased leakage into account when evaluating the results of radiation surveys around bunkers that are, or will be, used for IMRT. The recommended procedure of 15 applying a monitor-unit based workload correction factor to secondary barrier survey measurements, to account for this increased leakage when evaluating radiation survey measurements around IMRT bunkers, can lead to potentially-costly over estimation of the required barrier thickness. This study aims to provide initial guidance on the validity of reducing the value of the correction factor when applied to different radiation barriers (primary barriers, doors, maze walls and other walls) by 20 evaluating three different bunker designs. Methods Radiation survey measurements of primary, scattered and leakage radiation were obtained at each of five survey points around each of three different radiotherapy bunkers and the contribution of leakage to the total measured radiation dose at each point was evaluated. Measurements at each survey point were made with the linac gantry set to 12 equidistant positions from 0 to 330o, to 25 assess the effects of radiation beam direction on the results. Results For all three bunker designs, less than 0.5% of dose measured at and alongside the primary barriers, less than 25% of the dose measured outside the bunker doors and up to 100% of the dose measured outside other secondary barriers was found to be caused by linac head leakage. Conclusions Results of this study suggest that IMRT workload corrections are unnecessary, for 30 survey measurements made at and alongside primary barriers. Use of reduced IMRT workload correction factors is recommended when evaluating survey measurements around a bunker door, provided that a subset of the measurements used in this study are repeated for the bunker in question. Reduction of the correction factor for other secondary barrier survey measurements is not recommended unless the contribution from leakage is separetely evaluated.

Development of empirical equations for irradiance profile of a standard parabolic trough collector using Monte Carlo ray tracing technique

Irradiance profile around the receiver tube (RT) of a parabolic trough collector (PTC) is a key effect of optical performance that affects the overall energy performance of the collector. Thermal performance evaluation of the RT relies on the appropriate determination of the irradiance profile. This article explains a technique in which empirical equations were developed to calculate the local irradiance as a function of angular location of the RT of a standard PTC using a vigorously verified Monte Carlo ray tracing model. A large range of test conditions including daily normal insolation, spectral selective coatings and glass envelop conditions were selected from the published data by Dudley et al. [1] for the job. The R2 values of the equations are excellent that vary in between 0.9857 and 0.9999. Therefore, these equations can be used confidently to produce realistic non-uniform boundary heat flux profile around the RT at normal incidence for conjugate heat transfer analyses of the collector. Required values in the equations are daily normal insolation, and the spectral selective properties of the collector components. Since the equations are polynomial functions, data processing software can be employed to calculate the flux profile very easily and quickly. The ultimate goal of this research is to make the concentrating solar power technology cost competitive with conventional energy technology facilitating its ongoing research.

Controlled Monte Carlo data generation for statistical damage identification employing Mahalanobis squared distance

The use of Mahalanobis squared distance–based novelty detection in statistical damage identification has become increasingly popular in recent years. The merit of the Mahalanobis squared distance–based method is that it is simple and requires low computational effort to enable the use of a higher dimensional damage-sensitive feature, which is generally more sensitive to structural changes. Mahalanobis squared distance–based damage identification is also believed to be one of the most suitable methods for modern sensing systems such as wireless sensors. Although possessing such advantages, this method is rather strict with the input requirement as it assumes the training data to be multivariate normal, which is not always available particularly at an early monitoring stage. As a consequence, it may result in an ill-conditioned training model with erroneous novelty detection and damage identification outcomes. To date, there appears to be no study on how to systematically cope with such practical issues especially in the context of a statistical damage identification problem. To address this need, this article proposes a controlled data generation scheme, which is based upon the Monte Carlo simulation methodology with the addition of several controlling and evaluation tools to assess the condition of output data. By evaluating the convergence of the data condition indices, the proposed scheme is able to determine the optimal setups for the data generation process and subsequently avoid unnecessarily excessive data. The efficacy of this scheme is demonstrated via applications to a benchmark structure data in the field.

A sequential Monte Carlo framework for adaptive Bayesian model discrimination designs using mutual information

In this paper we present a unified sequential Monte Carlo (SMC) framework for performing sequential experimental design for discriminating between a set of models. The model discrimination utility that we advocate is fully Bayesian and based upon the mutual information. SMC provides a convenient way to estimate the mutual information. Our experience suggests that the approach works well on either a set of discrete or continuous models and outperforms other model discrimination approaches.

Using narrow beam profiles to quantify focal spot size, for accurate Monte Carlo simulations of SRS/SRT systems

This study investigates the variation of photon field penumbra shape with initial electron beam diameter, for very narrow beams. A Varian Millenium MLC (Varian Medical Systems, Palo Alto, USA) and a Brainlab m3 microMLC (Brainlab AB. Feldkirchen, Germany) were used, with one Varian iX linear accelerator, to produce fields that were (nominally) 0.20 cm across. Dose profiles for these fields were measured using radiochromic film and compared with the results of simulations completed using BEAMnrc and DOSXYZnrc, where the initial electron beam was set to FWHM = 0.02, 0.10, 0.12, 0.15, 0.20 and 0.50 cm. Increasing the electron-beam FWHM produced increasing occlusion of the photon source by the closely spaced collimator leaves and resulted in blurring of the simulated profile widths from 0.26 to 0.64 cm, for the MLC, from 0.12 to 0.43 cm, for the microMLC. Comparison with measurement data suggested that the electron spot size in the clinical linear accelerator was between FWHM = 0.10 and 0.15 cm, encompassing the result of our previous output-factor based work, which identified a FWHM of 0.12. Investigation of narrow-beam penumbra variation has been found to be a useful procedure, with results varying noticeably with linear accelerator spot size and allowing FWHM estimates obtained using other methods to be verified.

The influence of Monte Carlo source parameters on detector design and dose perturbation in small field dosimetry

To obtain accurate Monte Carlo simulations of small radiation fields, it is important model the initial source parameters (electron energy and spot size) accurately. However recent studies have shown that small field dosimetry correction factors are insensitive to these parameters. The aim of this work is to extend this concept to test if these parameters affect dose perturbations in general, which is important for detector design and calculating perturbation correction factors. The EGSnrc C++ user code cavity was used for all simulations. Varying amounts of air between 0 and 2 mm were deliberately introduced upstream to a diode and the dose perturbation caused by the air was quantified. These simulations were then repeated using a range of initial electron energies (5.5 to 7.0 MeV) and electron spot sizes (0.7 to 2.2 FWHM). The resultant dose perturbations were large. For example 2 mm of air caused a dose reduction of up to 31% when simulated with a 6 mm field size. However these values did not vary by more than 2 % when simulated across the full range of source parameters tested. If a detector is modified by the introduction of air, one can be confident that the response of the detector will be the same across all similar linear accelerators and the Monte Carlo modelling of each machine is not required.

Efficiency comparison of Markov Chain Monte Carlo simulation with subset simulation (MCMC/ss) to standard Monte Carlo Simulation (sMC) for extreme event scenarios

Standard Monte Carlo (sMC) simulation models have been widely used in AEC industry research to address system uncertainties. Although the benefits of probabilistic simulation analyses over deterministic methods are well documented, the sMC simulation technique is quite sensitive to the probability distributions of the input variables. This phenomenon becomes highly pronounced when the region of interest within the joint probability distribution (a function of the input variables) is small. In such cases, the standard Monte Carlo approach is often impractical from a computational standpoint. In this paper, a comparative analysis of standard Monte Carlo simulation to Markov Chain Monte Carlo with subset simulation (MCMC/ss) is presented. The MCMC/ss technique constitutes a more complex simulation method (relative to sMC), wherein a structured sampling algorithm is employed in place of completely randomized sampling. Consequently, gains in computational efficiency can be made. The two simulation methods are compared via theoretical case studies.

A simple method for EPID-based in vivo dosimetry for radiotherapy treatments of head-and-neck cancers

This study used a homogeneous water-equivalent model of an electronic portal imaging device (EPID), contoured as a structure in a radiotherapy treatment plan, to produce reference dose images for comparison with in vivo EPID dosimetry images. Head and neck treatments were chosen as the focus of this study, due to the heterogeneous anatomies involved and the consequent difficulty of rapidly obtaining reliable reference dose images by other means. A phantom approximating the size and heterogeneity of a typical neck, with a maximum radiological thickness of 8.5 cm, was constructed for use in this study. This phantom was CT scanned and a simple treatment including five square test fields and one off-axis IMRT field was planned. In order to allow the treatment planning system to calculate dose in a model EPID positioned a distance downstream from the phantom to achieve a source-to-detector distance (SDD) of 150 cm, the CT images were padded with air and the phantom’s “body” contour was extended to encompass the EPID contour. Comparison of dose images obtained from treatment planning calculations and experimental irradiations showed good agreement, with more than 90% of points in all fields passing a gamma evaluation, at γ (3%, 3mm )Similar agreement was achieved when the phantom was over-written with air in the treatment plan and removed from the experimental beam, suggesting that water EPID model at 150 cm SDD is capable of providing accurate reference images for comparison with clinical IMRT treatment images, for patient anatomies with radiological thicknesses ranging from 0 up to approximately 9 cm. This methodology therefore has the potential to be used for in vivo dosimetry during treatments to tissues in the neck as well as the oral and nasal cavities, in the head-and-neck region.

Monte Carlo simulations of small field output factors in solid phantoms

Introduction Total scatter factor (or output factor) in megavoltage photon dosimetry is a measure of relative dose relating a certain field size to a reference field size. The use of solid phantoms has been well established for output factor measurements, however to date these phantoms have not been tested with small fields. In this work, we evaluate the water equivalency of a number of solid phantoms for small field output factor measurements using the EGSnrc Monte Carlo code. Methods The following small square field sizes were simulated using BEAMnrc: 5, 6, 7, 8, 10 and 30 mm. Each simulated phantom geometry was created in DOSXYZnrc and consisted of a silicon diode (of length and width 1.5 mm and depth 0.5 mm) submersed in the phantom at a depth of 5 g/cm2. The source-to-detector distance was 100 cm for all simulations. The dose was scored in a single voxel at the location of the diode. Interaction probabilities and radiation transport parameters for each material were created using custom PEGS4 files. Results A comparison of the resultant output factors in the solid phantoms, compared to the same factors in a water phantom are shown in Fig. 1. The statistical uncertainty in each point was less than or equal to 0.4 %. The results in Fig. 1 show that the density of the phantoms affected the output factor results, with higher density materials (such as PMMA) resulting in higher output factors. Additionally, it was also calculated that scaling the depth for equivalent path length had negligible effect on the output factor results at these field sizes. Discussion and conclusions Electron stopping power and photon mass energy absorption change minimally with small field size [1]. Also, it can be seen from Fig. 1 that the difference from water decreases with increasing field size. Therefore, the most likely cause for the observed discrepancies in output factors is differing electron disequilibrium as a function of phantom density. When measuring small field output factors in a solid phantom, it is important that the density is very close to that of water.