Analysis of Cone-Beam CT using prior information

Treatment plans for conformal radiotherapy are based on an initial CT scan. The aim is to deliver the prescribed dose to the tumour, while minimising exposure to nearby organs. Recent advances make it possible to also obtain a Cone-Beam CT (CBCT) scan, once the patient has been positioned for treatment. A statistical model will be developed to compare these CBCT scans with the initial CT scan. Changes in the size, shape and position of the tumour and organs will be detected and quantified. Some progress has already been made in segmentation of prostate CBCT scans [1],[2],[3]. However, none of the existing approaches have taken full advantage of the prior information that is available. The planning CT scan is expertly annotated with contours of the tumour and nearby sensitive objects. This data is specific to the individual patient and can be viewed as a snapshot of spatial information at a point in time. There is an abundance of studies in the radiotherapy literature that describe the amount of variation in the relevant organs between treatments. The findings from these studies can form a basis for estimating the degree of uncertainty. All of this information can be incorporated as an informative prior into a Bayesian statistical model. This model will be developed using scans of CT phantoms, which are objects with known geometry. Thus, the accuracy of the model can be evaluated objectively. This will also enable comparison between alternative models.

Bayesian algorithms with applications

Advances in algorithms for approximate sampling from a multivariable target function have led to solutions to challenging statistical inference problems that would otherwise not be considered by the applied scientist. Such sampling algorithms are particularly relevant to Bayesian statistics, since the target function is the posterior distribution of the unobservables given the observables. In this thesis we develop, adapt and apply Bayesian algorithms, whilst addressing substantive applied problems in biology and medicine as well as other applications. For an increasing number of high-impact research problems, the primary models of interest are often sufficiently complex that the likelihood function is computationally intractable. Rather than discard these models in favour of inferior alternatives, a class of Bayesian "likelihoodfree" techniques (often termed approximate Bayesian computation (ABC)) has emerged in the last few years, which avoids direct likelihood computation through repeated sampling of data from the model and comparing observed and simulated summary statistics. In Part I of this thesis we utilise sequential Monte Carlo (SMC) methodology to develop new algorithms for ABC that are more efficient in terms of the number of model simulations required and are almost black-box since very little algorithmic tuning is required. In addition, we address the issue of deriving appropriate summary statistics to use within ABC via a goodness-of-fit statistic and indirect inference. Another important problem in statistics is the design of experiments. That is, how one should select the values of the controllable variables in order to achieve some design goal. The presences of parameter and/or model uncertainty are computational obstacles when designing experiments but can lead to inefficient designs if not accounted for correctly. The Bayesian framework accommodates such uncertainties in a coherent way. If the amount of uncertainty is substantial, it can be of interest to perform adaptive designs in order to accrue information to make better decisions about future design points. This is of particular interest if the data can be collected sequentially. In a sense, the current posterior distribution becomes the new prior distribution for the next design decision. Part II of this thesis creates new algorithms for Bayesian sequential design to accommodate parameter and model uncertainty using SMC. The algorithms are substantially faster than previous approaches allowing the simulation properties of various design utilities to be investigated in a more timely manner. Furthermore the approach offers convenient estimation of Bayesian utilities and other quantities that are particularly relevant in the presence of model uncertainty. Finally, Part III of this thesis tackles a substantive medical problem. A neurological disorder known as motor neuron disease (MND) progressively causes motor neurons to no longer have the ability to innervate the muscle fibres, causing the muscles to eventually waste away. When this occurs the motor unit effectively ‘dies’. There is no cure for MND, and fatality often results from a lack of muscle strength to breathe. The prognosis for many forms of MND (particularly amyotrophic lateral sclerosis (ALS)) is particularly poor, with patients usually only surviving a small number of years after the initial onset of disease. Measuring the progress of diseases of the motor units, such as ALS, is a challenge for clinical neurologists. Motor unit number estimation (MUNE) is an attempt to directly assess underlying motor unit loss rather than indirect techniques such as muscle strength assessment, which generally is unable to detect progressions due to the body’s natural attempts at compensation. Part III of this thesis builds upon a previous Bayesian technique, which develops a sophisticated statistical model that takes into account physiological information about motor unit activation and various sources of uncertainties. More specifically, we develop a more reliable MUNE method by applying marginalisation over latent variables in order to improve the performance of a previously developed reversible jump Markov chain Monte Carlo sampler. We make other subtle changes to the model and algorithm to improve the robustness of the approach.

Prediction of IBD based on population history for fine gene mapping

A novel multiple regression method (RM) is developed to predict identity-by-descent probabilities at a locus L (IBDL), among individuals without pedigree, given information on surrounding markers and population history. These IBDL probabilities are a function of the increase in linkage disequilibrium (LD) generated by drift in a homogeneous population over generations. Three parameters are sufficient to describe population history: effective population size (Ne), number of generations since foundation (T), and marker allele frequencies among founders (p). IBD L are used in a simulation study to map a quantitative trait locus (QTL) via variance component estimation. RM is compared to a coalescent method (CM) in terms of power and robustness of QTL detection. Differences between RM and CM are small but significant. For example, RM is more powerful than CM in dioecious populations, but not in monoecious populations. Moreover, RM is more robust than CM when marker phases are unknown or when there is complete LD among founders or Ne is wrong, and less robust when p is wrong. CM utilises all marker haplotype information, whereas RM utilises information contained in each individual marker and all possible marker pairs but not in higher order interactions. RM consists of a family of models encompassing four different population structures, and two ways of using marker information, which contrasts with the single model that must cater for all possible evolutionary scenarios in CM.

On the prediction of simultaneous inbreeding coefficients at multiple loci

A new deterministic method for predicting simultaneous inbreeding coefficients at three and four loci is presented. The method involves calculating the conditional probability of IBD (identical by descent) at one locus given IBD at other loci, and multiplying this probability by the prior probability of the latter loci being simultaneously IBD. The conditional probability is obtained applying a novel regression model, and the prior probability from the theory of digenic measures of Weir and Cockerham. The model was validated for a finite monoecious population mating at random, with a constant effective population size, and with or without selfing, and also for an infinite population with a constant intermediate proportion of selfing. We assumed discrete generations. Deterministic predictions were very accurate when compared with simulation results, and robust to alternative forms of implementation. These simultaneous inbreeding coefficients were more sensitive to changes in effective population size than in marker spacing. Extensions to predict simultaneous inbreeding coefficients at more than four loci are now possible.

A Bayesian approach to the determination of ignition delay

A novel in-cylinder pressure method for determining ignition delay has been proposed and demonstrated. This method proposes a new Bayesian statistical model to resolve the start of combustion, defined as being the point at which the band-pass in-cylinder pressure deviates from background noise and the combustion resonance begins. Further, it is demonstrated that this method is still accurate in situations where there is noise present. The start of combustion can be resolved for each cycle without the need for ad hoc methods such as cycle averaging. Therefore, this method allows for analysis of consecutive cycles and inter-cycle variability studies. Ignition delay obtained by this method and by the net rate of heat release have been shown to give good agreement. However, the use of combustion resonance to determine the start of combustion is preferable over the net rate of heat release method because it does not rely on knowledge of heat losses and will still function accurately in the presence of noise. Results for a six-cylinder turbo-charged common-rail diesel engine run with neat diesel fuel at full, three quarters and half load have been presented. Under these conditions the ignition delay was shown to increase as the load was decreased with a significant increase in ignition delay at half load, when compared with three quarter and full loads.

Combining elicited expert knowledge into Bayesian priors

Information that is elicited from experts can be treated as `data', so can be analysed using a Bayesian statistical model, to formulate a prior model. Typically methods for encoding a single expert's knowledge have been parametric, constrained by the extent of an expert's knowledge and energy regarding a target parameter. Interestingly these methods have often been deterministic, in that all elicited information is treated at `face value', without error. Here we sought a parametric and statistical approach for encoding assessments from multiple experts. Our recent work proposed and demonstrated the use of a flexible hierarchical model for this purpose. In contrast to previous mathematical approaches like linear or geometric pooling, our new approach accounts for several sources of variation: elicitation error, encoding error and expert diversity. Of interest are the practical, mathematical and philosophical interpretations of this form of hierarchical pooling (which is both statistical and parametric), and how it fits within the subjective Bayesian paradigm. Case studies from a bioassay and project management (on PhDs) are used to illustrate the approach.

Segmentation of cone-beam CT using a hidden Markov random field with informative priors

Purpose: Flat-detector, cone-beam computed tomography (CBCT) has enormous potential to improve the accuracy of treatment delivery in image-guided radiotherapy (IGRT). To assist radiotherapists in interpreting these images, we use a Bayesian statistical model to label each voxel according to its tissue type. Methods: The rich sources of prior information in IGRT are incorporated into a hidden Markov random field (MRF) model of the 3D image lattice. Tissue densities in the reference CT scan are estimated using inverse regression and then rescaled to approximate the corresponding CBCT intensity values. The treatment planning contours are combined with published studies of physiological variability to produce a spatial prior distribution for changes in the size, shape and position of the tumour volume and organs at risk (OAR). The voxel labels are estimated using the iterated conditional modes (ICM) algorithm. Results: The accuracy of the method has been evaluated using 27 CBCT scans of an electron density phantom (CIRS, Inc. model 062). The mean voxel-wise misclassification rate was 6.2%, with Dice similarity coefficient of 0.73 for liver, muscle, breast and adipose tissue. Conclusions: By incorporating prior information, we are able to successfully segment CBCT images. This could be a viable approach for automated, online image analysis in radiotherapy.

Quantitative genetic parameter estimates for body and carcass traits in a cultured stock of giant freshwater prawn (Macrobrachium rosenbergii) selected for harvest weight in Vietnam

We estimated the heritability and correlations between body and carcass weight traits in a cultured stock of giant freshwater prawn (GFP) (Macrobrachium rosenbergii) selected for harvest body weight in Vietnam. The data set consisted of 18,387 body and 1,730 carcass records, as well as full pedigree information collected over four generations. Variance and covariance components were estimated by restricted maximum likelihood fitting a multi-trait animal model. Across generations, estimates of heritability for body and carcass weight traits were moderate and ranged from 0.14 to 0.19 and 0.17 to 0.21, respectively. Body trait heritabilities estimated for females were significantly higher than for males whereas carcass weight trait heritabilities estimated for females and males were not significantly different (P>. 0.05). Maternal effects for body traits accounted for 4 to 5% of the total variance and were greater in females than in males. Genetic correlations among body traits were generally high in the mixed sexes. Genetic correlations between body and carcass weight traits were also high. Although some issues remain regarding the best statistical model to be fitted to GFP data, our results suggest that selection for high harvest body weight based on breeding values estimated by fitting an animal model to the data can significantly improve mean body and carcass weight in GFP.

Segmentation of cone-beam CT using a hidden Markov random field with informative priors

Cone-beam computed tomography (CBCT) has enormous potential to improve the accuracy of treatment delivery in image-guided radiotherapy (IGRT). To assist radiotherapists in interpreting these images, we use a Bayesian statistical model to label each voxel according to its tissue type. The rich sources of prior information in IGRT are incorporated into a hidden Markov random field model of the 3D image lattice. Tissue densities in the reference CT scan are estimated using inverse regression and then rescaled to approximate the corresponding CBCT intensity values. The treatment planning contours are combined with published studies of physiological variability to produce a spatial prior distribution for changes in the size, shape and position of the tumour volume and organs at risk. The voxel labels are estimated using iterated conditional modes. The accuracy of the method has been evaluated using 27 CBCT scans of an electron density phantom. The mean voxel-wise misclassification rate was 6.2\%, with Dice similarity coefficient of 0.73 for liver, muscle, breast and adipose tissue. By incorporating prior information, we are able to successfully segment CBCT images. This could be a viable approach for automated, online image analysis in radiotherapy.

Estimating present day extreme total water level exceedance probabilities around the Australian coastline

The occurrence of extreme water level events along low-lying, highly populated and/or developed coastlines can lead to devastating impacts on coastal infrastructure. Therefore it is very important that the probabilities of extreme water levels are accurately evaluated to inform flood and coastal management and for future planning. The aim of this study was to provide estimates of present day extreme total water level exceedance probabilities around the whole coastline of Australia, arising from combinations of mean sea level, astronomical tide and storm surges generated by both extra-tropical and tropical storms, but exclusive of surface gravity waves. The study has been undertaken in two main stages. In the first stage, a high-resolution (~10 km along the coast) hydrodynamic depth averaged model has been configured for the whole coastline of Australia using the Danish Hydraulics Institute’s Mike21 modelling suite of tools. The model has been forced with astronomical tidal levels, derived from the TPX07.2 global tidal model, and meteorological fields, from the US National Center for Environmental Prediction’s global reanalysis, to generate a 61-year (1949 to 2009) hindcast of water levels. This model output has been validated against measurements from 30 tide gauge sites around Australia with long records. At each of the model grid points located around the coast, time series of annual maxima and the several highest water levels for each year were derived from the multi-decadal water level hindcast and have been fitted to extreme value distributions to estimate exceedance probabilities. Stage 1 provided a reliable estimate of the present day total water level exceedance probabilities around southern Australia, which is mainly impacted by extra-tropical storms. However, as the meteorological fields used to force the hydrodynamic model only weakly include the effects of tropical cyclones the resultant water levels exceedance probabilities were underestimated around western, northern and north-eastern Australia at higher return periods. Even if the resolution of the meteorological forcing was adequate to represent tropical cyclone-induced surges, multi-decadal periods yielded insufficient instances of tropical cyclones to enable the use of traditional extreme value extrapolation techniques. Therefore, in the second stage of the study, a statistical model of tropical cyclone tracks and central pressures was developed using histroic observations. This model was then used to generate synthetic events that represented 10,000 years of cyclone activity for the Australia region, with characteristics based on the observed tropical cyclones over the last ~40 years. Wind and pressure fields, derived from these synthetic events using analytical profile models, were used to drive the hydrodynamic model to predict the associated storm surge response. A random time period was chosen, during the tropical cyclone season, and astronomical tidal forcing for this period was included to account for non-linear interactions between the tidal and surge components. For each model grid point around the coast, annual maximum total levels for these synthetic events were calculated and these were used to estimate exceedance probabilities. The exceedance probabilities from stages 1 and 2 were then combined to provide a single estimate of present day extreme water level probabilities around the whole coastline of Australia.

A review of modern computational algorithms for Bayesian optimal design

Bayesian experimental design is a fast growing area of research with many real-world applications. As computational power has increased over the years, so has the development of simulation-based design methods, which involve a number of algorithms, such as Markov chain Monte Carlo, sequential Monte Carlo and approximate Bayes methods, facilitating more complex design problems to be solved. The Bayesian framework provides a unified approach for incorporating prior information and/or uncertainties regarding the statistical model with a utility function which describes the experimental aims. In this paper, we provide a general overview on the concepts involved in Bayesian experimental design, and focus on describing some of the more commonly used Bayesian utility functions and methods for their estimation, as well as a number of algorithms that are used to search over the design space to find the Bayesian optimal design. We also discuss other computational strategies for further research in Bayesian optimal design.

Burglar target selection : a cross-national comparison

Objectives This study builds on research undertaken by Bernasco and Nieuwbeerta and explores the generalizability of a theoretically derived offender target selection model in three cross-national study regions. Methods Taking a discrete spatial choice approach, we estimate the impact of both environment- and offender-level factors on residential burglary placement in the Netherlands, the United Kingdom, and Australia. Combining cleared burglary data from all study regions in a single statistical model, we make statistical comparisons between environments. Results In all three study regions, the likelihood an offender selects an area for burglary is positively influenced by proximity to their home, the proportion of easily accessible targets, and the total number of targets available. Furthermore, in two of the three study regions, juvenile offenders under the legal driving age are significantly more influenced by target proximity than adult offenders. Post hoc tests indicate the magnitudes of these impacts vary significantly between study regions. Conclusions While burglary target selection strategies are consistent with opportunity-based explanations of offending, the impact of environmental context is significant. As such, the approach undertaken in combining observations from multiple study regions may aid criminology scholars in assessing the generalizability of observed findings across multiple environments.

Meta-analysis of prevalence

Meta-analysis is a method to obtain a weighted average of results from various studies. In addition to pooling effect sizes, meta-analysis can also be used to estimate disease frequencies, such as incidence and prevalence. In this article we present methods for the meta-analysis of prevalence. We discuss the logit and double arcsine transformations to stabilise the variance. We note the special situation of multiple category prevalence, and propose solutions to the problems that arise. We describe the implementation of these methods in the MetaXL software, and present a simulation study and the example of multiple sclerosis from the Global Burden of Disease 2010 project. We conclude that the double arcsine transformation is preferred over the logit, and that the MetaXL implementation of multiple category prevalence is an improvement in the methodology of the meta-analysis of prevalence.

Initial burden of disease estimates for South Africa, 2000

BACKGROUND This paper describes the first national burden of disease study for South Africa. The main focus is the burden due to premature mortality, i.e. years of life lost (YLLs). In addition, estimates of the burden contributed by morbidity, i.e. the years lived with disability (YLDs), are obtained to calculate disability-adjusted life years (DALYs); and the impact of AIDS on premature mortality in the year 2010 is assessed. METHOD Owing to the rapid mortality transition and the lack of timely data, a modelling approach has been adopted. The total mortality for the year 2000 is estimated using a demographic and AIDS model. The non-AIDS cause-of-death profile is estimated using three sources of data: Statistics South Africa, the National Department of Home Affairs, and the National Injury Mortality Surveillance System. A ratio method is used to estimate the YLDs from the YLL estimates. RESULTS The top single cause of mortality burden was HIV/AIDS followed by homicide, tuberculosis, road traffic accidents and diarrhoea. HIV/AIDS accounted for 38% of total YLLs, which is proportionately higher for females (47%) than for males (33%). Pre-transitional diseases, usually associated with poverty and underdevelopment, accounted for 25%, non-communicable diseases 21% and injuries 16% of YLLs. The DALY estimates highlight the fact that mortality alone underestimates the burden of disease, especially with regard to unintentional injuries, respiratory disease, and nervous system, mental and sense organ disorders. The impact of HIV/AIDS is expected to more than double the burden of premature mortality by the year 2010. CONCLUSION This study has drawn together data from a range of sources to develop coherent estimates of premature mortality by cause. South Africa is experiencing a quadruple burden of disease comprising the pre-transitional diseases, the emerging chronic diseases, injuries, and HIV/AIDS. Unless interventions that reduce morbidity and delay morbidity become widely available, the burden due to HIV/AIDS can be expected to grow very rapidly in the next few years. An improved base of information is needed to assess the morbidity impact more accurately.

Global variation in the prevalence and incidence of major depressive disorder : a systematic review of the epidemiological literature

Background Summarizing the epidemiology of major depressive disorder (MDD) at a global level is complicated by significant heterogeneity in the data. The aim of this study is to present a global summary of the prevalence and incidence of MDD, accounting for sources of bias, and dealing with heterogeneity. Findings are informing MDD burden quantification in the Global Burden of Disease (GBD) 2010 Study. Method A systematic review of prevalence and incidence of MDD was undertaken. Electronic databases Medline, PsycINFO and EMBASE were searched. Community-representative studies adhering to suitable diagnostic nomenclature were included. A meta-regression was conducted to explore sources of heterogeneity in prevalence and guide the stratification of data in a meta-analysis. Results The literature search identified 116 prevalence and four incidence studies. Prevalence period, sex, year of study, depression subtype, survey instrument, age and region were significant determinants of prevalence, explaining 57.7% of the variability between studies. The global point prevalence of MDD, adjusting for methodological differences, was 4.7% (4.4–5.0%). The pooled annual incidence was 3.0% (2.4–3.8%), clearly at odds with the pooled prevalence estimates and the previously reported average duration of 30 weeks for an episode of MDD. Conclusions Our findings provide a comprehensive and up-to-date profile of the prevalence of MDD globally. Region and study methodology influenced the prevalence of MDD. This needs to be considered in the GBD 2010 study and in investigations into the ecological determinants of MDD. Good-quality estimates from low-/middle-income countries were sparse. More accurate data on incidence are also required.