Paul A. Bromiley

Robust and Accurate Shape Model Matching Using Random Forest Regression-Voting

A widely used approach for locating points on deformable objects in images is to generate feature response images for each point, and then to fit a shape model to these response images. We demonstrate that Random Forest regression-voting can be used to generate high quality response images quickly. Rather than using a generative or a discriminative model to evaluate each pixel, a regressor is used to cast votes for the optimal position of each point. We show that this leads to fast and accurate shape model matching when applied in the Constrained Local Model framework. We evaluate the technique in detail, and compare it with a range of commonly used alternatives across application areas: the annotation of the joints of the hands in radiographs and the detection of feature points in facial images. We show that our approach outperforms alternative techniques, achieving what we believe to be the most accurate results yet published for hand joint annotation and state-of-the-art performance for facial feature point detection.

Real-time Body Tracking Using a Gaussian Process Latent Variable Model

In this paper, we present a tracking framework for capturing articulated human motions in real-time, without the need for attaching markers onto the subjects body. This is achieved by first obtaining a low dimensional representation of the training motion data, using a nonlinear dimensionality reduction technique called back-constrained GPLVM. A prior dynamics model is then learnt from this low dimensional representation by partitioning the motion sequences into elementary movements using an unsupervised EM clustering algorithm. The temporal dependencies between these elementary movements are efficiently captured by a Variable Length Markov Model. The learnt dynamics model is used to bias the propagation of candidate pose feature vectors in the low dimensional space. By combining this with an efficient volumetric reconstruction algorithm, our framework can quickly evaluate each candidate pose against image evidence captured from multiple views. We present results that show our system can accurately track complex structured activities such as ballet dancing in real-time.

Non-Parametric Image Subtraction using Grey Level Scattergrams

Image subtraction is used in many areas of machine vision to identify small changes between equivalent pairs of images. Often only a small subset of the differences will be of interest. In motion analysis only those differences caused by motion are important, and differences due to other sources only serve to complicate interpretation. Simple image subtraction detects all differences regardless of their source, and is therefore problematic to use. Superior techniques, analogous to standard statistical tests, can isolate localised differences due to motion from global differences due, for example, to illumination changes. Four such techniques are described. In particular, we introduce a new non-parametric statistical measure which allows a direct probabilistic interpretation of image differences. We expect this to be applicable to a wide range of image formation processes. Its application to medical images is discussed.

A fast, model-independent method for cerebral cortical thickness estimation using MRI

Several algorithms for measuring the cortical thickness in the human brain from MR image volumes have been described in the literature, the majority of which rely on fitting deformable models to the inner and outer cortical surfaces. However, the constraints applied during the model fitting process in order to enforce spherical topology and to fit the outer cortical surface in narrow sulci, where the cerebrospinal fluid (CSF) channel may be obscured by partial voluming, may introduce bias in some circumstances, and greatly increase the processor time required. In this paper we describe an alternative, voxel based technique that measures the cortical thickness using inversion recovery anatomical MR images. Grey matter, white matter and CSF are identified through segmentation, and edge detection is used to identify the boundaries between these tissues. The cortical thickness is then measured along the local 3D surface normal at every voxel on the inner cortical surface. The method was applied to 119 normal volunteers, and validated through extensive comparisons with published measurements of both cortical thickness and rate of thickness change with age. We conclude that the proposed technique is generally faster than deformable model-based alternatives, and free from the possibility of model bias, but suffers no reduction in accuracy. In particular, it will be applicable in data sets showing severe cortical atrophy, where thinning of the gyri leads to points of high curvature, and so the fitting of deformable models is problematic.

Endoscopic Third Ventriculostomy: Predicting Outcome with Phase-Contrast MR Imaging

PURPOSE To determine whether phase-contrast magnetic resonance (MR) imaging measurements of preoperative cerebral blood and cerebrospinal fluid (CSF) hydrodynamics can be used as a biomarker of response to endoscopic third ventriculostomy (ETV). MATERIALS AND METHODS Approval from the local research ethics committee and written informed consent were obtained for this prospective study. Thirteen patients (six female patients, seven male patients; median age, 43 years) with chronic obstructive hydrocephalus, 12 of whom went on to undergo ETV, were imaged with phase-contrast MR imaging at 1.5 T to determine rates of total cerebral blood flow (CBF) and ventriculostomy defect, foramen magnum (FM), and cerebral aqueduct CSF flow. Ten control subjects (10 men; median age, 37 years) were similarly imaged. Correlations between measured values were assessed by means of Pearson correlation coefficients. Measurements were compared between groups with a Mann-Whitney test, and measurements before and after surgical intervention were compared with a Wilcoxon test for paired samples. RESULTS Rates of CBF (356 mL . min(-1) +/- 73 [standard deviation] vs 518 mL . min(-1) +/- 79, P < .001) and CSF flow in the FM (17.62 mL . min(-1) +/- 13.12 vs 36.35 mL . min(-1) +/- 8, P < .05) were significantly lower in patients than in control subjects. CONCLUSION ETV induces changes in brain volume and CBF that can be predicted by using simple metrics. These pilot results support a formal trial of these techniques in a larger prospective study.

TINA manual landmarking tool: software for the precise digitization of 3D landmarks

BackgroundInterest in the placing of landmarks and subsequent morphometric analyses of shape for 3D data has increased with the increasing accessibility of computed tomography (CT) scanners. However, current computer programs for this task suffer from various practical drawbacks. We present here a free software tool that overcomes many of these problems.ResultsThe TINA Manual Landmarking Tool was developed for the digitization of 3D data sets. It enables the generation of a modifiable 3D volume rendering display plus matching orthogonal 2D cross-sections from DICOM files. The object can be rotated and axes defined and fixed. Predefined lists of landmarks can be loaded and the landmarks identified within any of the representations. Output files are stored in various established formats, depending on the preferred evaluation software.ConclusionsThe software tool presented here provides several options facilitating the placing of landmarks on 3D objects, including volume rendering from DICOM files, definition and fixation of meaningful axes, easy import, placement, control, and export of landmarks, and handling of large datasets. The TINA Manual Landmark Tool runs under Linux and can be obtained for free from http://www.tina-vision.net/tarballs/.

Bayesian and non-Bayesian probabilistic models for medical image analysis

Abstract Bayesian approaches to data analysis are popular in machine vision, and yet the main advantage of Bayes theory, the ability to incorporate prior knowledge in the form of the prior probabilities, may lead to problems in some quantitative tasks. In this paper we demonstrate examples of Bayesian and non-Bayesian techniques from the area of magnetic resonance image (MRI) analysis. Issues raised by these examples are used to illustrate difficulties in Bayesian methods and to motivate an approach based on frequentist methods. We believe this approach to be more suited to quantitative data analysis, and provide a general theory for the use of these methods in learning (Bayes risk) systems and for data fusion. Proofs are given for the more novel aspects of the theory. We conclude with a discussion of the strengths and weaknesses, and the fundamental suitability, of Bayesian and non-Bayesian approaches for MRI analysis in particular, and for machine vision systems in general.

Semi-automatic landmark point annotation for geometric morphometrics

BackgroundIn previous work, the authors described a software package for the digitisation of 3D landmarks for use in geometric morphometrics. In this paper, we describe extensions to this software that allow semi-automatic localisation of 3D landmarks, given a database of manually annotated training images. Multi-stage registration was applied to align image patches from the database to a query image, and the results from multiple database images were combined using an array-based voting scheme. The software automatically highlights points that have been located with low confidence, allowing manual correction.ResultsEvaluation was performed on micro-CT images of rodent skulls for which two independent sets of manual landmark annotations had been performed. This allowed assessment of landmark accuracy in terms of both the distance between manual and automatic annotations, and the repeatability of manual and automatic annotation. Automatic annotation attained accuracies equivalent to those achievable through manual annotation by an expert for 87.5% of the points, with significantly higher repeatability.ConclusionsWhilst user input was required to produce the training data and in a final error correction stage, the software was capable of reducing the number of manual annotations required in a typical landmark identification process using 3D data by a factor of ten, potentially allowing much larger data sets to be annotated and thus increasing the statistical power of the results from subsequent processing e.g. Procrustes/principal component analysis. The software is freely available, under the GNU General Public Licence, from our web-site (www.tina-vision.net).

Localisation of Vertebrae on DXA Images using Constrained Local Models with Random Forest Regression Voting

Fractures associated with osteoporosis are a significant public health risk, and one that is likely to increase with an ageing population. However, many osteoporotic vertebral fractures present on images do not come to clinical attention or lead to preventative treatment. Furthermore, vertebral fracture assessment (VFA) typically depends on subjective judgement by a radiologist. The potential utility of computer-aided VFA systems is therefore considerable. Previous work has shown that Active Appearance Models (AAMs) give accurate results when locating landmarks on vertebra in DXA images, but can give poor fits in a substantial subset of examples, particularly the more severe fractures. Here we evaluate Random Forest Regression Voting Constrained Local Models (RFRV-CLMs) for this task and show that, while they lead to slightly poorer median errors than AAMs, they are much more robust, reducing the proportion of fit failures by 68\%. They are thus more suitable for use in computer-aided VFA systems.

Biomechanical simulation of atrophy in MR images

Progressive cerebral atrophy is a physical component of the most common forms of dementia - Alzheimers disease, vascular dementia, Lewy-Body disease and fronto-temporal dementia. We propose a phenomenological simulation of atrophy in MR images that provides gold-standard data; the origin and rate of progression of atrophy can be controlled and the resultant remodelling of brain structures is known. We simulate diffuse global atrophic change by generating global volumetric change in a physically realistic biomechanical model of the human brain. Thermal loads are applied to either single, or multiple, tissue types within the brain to drive tissue expansion or contraction. Mechanical readjustment is modelled using finite element methods (FEM). In this preliminary work we apply these techniques to the MNI brainweb phantom to produce new images exhibiting global diffuse atrophy. We compare the applied atrophy with that measured from the images using an established quantitative technique. Early results are encouraging and suggest that the model can be extended and used for validation of atrophy measurement techniques and non-rigid image registration, and for understanding the effect of atrophy on brain shape.