Andrew H Gee

University of Cambridge, Cambridge, England, United Kingdom

Are you Andrew H Gee?

Claim your profile

Publications (158)209.72 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: MicroabstractThe aim of this study was to examine the effects of denosumab versus placebo on cortical bone using clinical CT. We used cortical bone mapping to study the hips of women with osteoporosis treated during the FREEDOM clinical trial. We found that cortical mass surface density and thickness increased rapidly during denosumab therapy, particularly in the hip trochanteric region.Women with osteoporosis treated for 36 months with twice-yearly injections of denosumab sustained fewer hip fractures compared with placebo. Treatment might improve femoral bone at locations where fractures typically occur. To test this hypothesis, we used 3D cortical bone mapping of postmenopausal women with osteoporosis, to investigate the timing and precise location of denosumab versus placebo effects in the hips.We analysed clinical CT scans from 80 female participants in FREEDOM, a randomised trial, wherein half of the study participants received subcutaneous denosumab 60mg twice-yearly and the others received placebo. Cortical 3D bone thickness maps of both hips were created from scans at baseline, 12, 24 and 36 months. Cortical mass surface density maps were also created for each visit. After registration of each bone to an average femur shape model followed by statistical parametric mapping, we visualised and quantified statistically significant treatment effects. The technique allowed us to pinpoint systematic differences between denosumab and control, and display the results on a 3D average femur model.Denosumab treatment led to an increase in femoral cortical mass surface density and thickness, already evident by the third injection (12 months). Overall, treatment with denosumab increased femoral cortical mass surface density by 5.4% over three years. One third of the increase came from increasing cortical density, and two thirds from increasing cortical thickness, relative to placebo. After 36 months, cortical mass surface density and thickness had increased by up to 12%, at key locations such as the lateral femoral trochanter, versus placebo. Most of the femoral cortex displayed a statistically significant relative difference by 36 months.Osteoporotic cortical bone responds rapidly to denosumab therapy, particularly in the hip trochanteric region. This mechanism may be involved in the robust decrease in hip fractures seen in denosumab treated women at increased risk of fracture. © 2014 American Society for Bone and Mineral Research
    Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research 08/2014; · 6.04 Impact Factor
  • Andrew H Gee, Graham M Treece
    [Show abstract] [Hide abstract]
    ABSTRACT: Spatial normalisation is a key element of statistical parametric mapping and related techniques for analysing cohort statistics on voxel arrays and surfaces. The normalisation process involves aligning each individual specimen to a template using some sort of registration algorithm. Any misregistration will result in data being mapped onto the template at the wrong location. At best, this will introduce spatial imprecision into the subsequent statistical analysis. At worst, when the misregistration varies systematically with a covariate of interest, it may lead to false statistical inference. Since misregistration generally depends on the specimen's shape, we investigate here the effect of allowing for shape as a confound in the statistical analysis, with shape represented by the dominant modes of variation observed in the cohort. In a series of experiments on synthetic surface data, we demonstrate how allowing for shape can reveal true effects that were previously masked by systematic misregistration, and also guard against misinterpreting systematic misregistration as a true effect. We introduce some heuristics for disentangling misregistration effects from true effects, and demonstrate the approach's practical utility in a case study of the cortical bone distribution in 268 human femurs.
    Medical image analysis 01/2014; 18(2):385-393. · 3.09 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: An ultrasound image is created from backscattered echoes originating from both diffuse and directional scattering. It is potentially useful to separate these two components for the purpose of tissue characterization. This article presents several models for visualization of scattering fields on 3-dimensional (3D) ultrasound imaging. By scanning the same anatomy from multiple directions, we can observe the variation of specular intensity as a function of the viewing angle. This article considers two models for estimating the diffuse and specular components of the backscattered intensity: a modification of the well-known Phong reflection model and an existing exponential model. We examine 2-dimensional implementations and also propose novel 3D extensions of these models in which the probe is not constrained to rotate within a plane. Both simulation and experimental results show that improved performance can be achieved with 3D models.
    Journal of ultrasound in medicine: official journal of the American Institute of Ultrasound in Medicine 04/2013; 32(4):699-714. · 1.40 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Hip fracture is the leading cause of acute orthopaedic hospital admission amongst the elderly, with around a third of patients not surviving one year post-fracture. Although various preventative therapies are available, patient selection is difficult. The current state-of-the-art risk assessment tool (FRAX) ignores focal structural defects, such as cortical bone thinning, a critical component in characterizing hip fragility. Cortical thickness can be measured using CT, but this is expensive and involves a significant radiation dose. Instead, Dual-Energy X-ray Absorptiometry (DXA) is currently the preferred imaging modality for assessing hip fracture risk and is used routinely in clinical practice. Our ambition is to develop a tool to measure cortical thickness using multi-view DXA instead of CT. In this initial study, we work with digitally reconstructed radiographs (DRRs) derived from CT data as a surrogate for DXA scans: this enables us to compare directly the thickness estimates with the gold standard CT results. Our approach involves a model-based femoral shape reconstruction followed by a data-driven algorithm to extract numerous cortical thickness point estimates. In a series of experiments on the shaft and trochanteric regions of 48 proximal femurs, we validated our algorithm and established its performance limits using 20 views in the range 0°-171°: estimation errors were 0:19 +/- 0:53mm (mean +/- one standard deviation). In a more clinically viable protocol using four views in the range 0°-51°, where no other bony structures obstruct the projection of the femur, measurement errors were -0:07 +/- 0:79 mm.
    Proc SPIE 02/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Most quasi-static ultrasound elastography methods image only the axial strain, derived from displacements measured in the direction of ultrasound propagation. In other directions, the beam lacks high resolution phase information and displacement estimation is therefore less precise. However, these estimates can be improved by steering the ultrasound beam through multiple angles and combining displacements measured along the different beam directions. Previously, beamsteering has only considered the 2D case to improve the lateral displacement estimates. In this paper, we extend this to 3D using a simulated 2D array to steer both laterally and elevationally in order to estimate the full 3D displacement vector over a volume. The method is tested on simulated and phantom data using a simulated 6-10MHz array, and the precision of displacement estimation is measured with and without beamsteering. In simulations, we found a statistically significant improvement in the precision of lateral and elevational displacement estimates: lateral precision 35.69μm unsteered, 3.70μm steered; elevational precision 38.67μm unsteered, 3.64μm steered. Similar results were found in the phantom data: lateral precision 26.51μm unsteered, 5.78μm steered; elevational precision 28.92μm unsteered, 11.87μm steered. We conclude that volumetric 3D beamsteering improves the precision of lateral and elevational displacement estimates.
    Ultrasonics 11/2012; · 2.03 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Objective This work is concerned with the creation of three-dimensional (3D) extended-field-of-view ultrasound from a set of volumes acquired using a mechanically swept 3D probe. 3D volumes of ultrasound data can be registered by attaching a position sensor to the probe; this can be an inconvenience in a clinical setting. A position sensor can also cause some misalignment due to patient movement and respiratory motion. We propose a combination of three-degrees-of-freedom image registration and an unobtrusively integrated inertial sensor for measuring orientation. The aim of this research is to produce a reliable and portable ultrasound system that is able to register 3D volumes quickly, making it suitable for clinical use. Method As part of a feasibility study we recruited 28 pregnant females attending for routine obstetric scans to undergo 3D extended-field-of-view ultrasound. A total of 49 data sets were recorded. Each registered data set was assessed for correct alignment of each volume by two independent observers. Results In 77-83% of the data sets more than four consecutive volumes registered. The successful registration relies on good overlap between volumes and is adversely affected by advancing gestational age and foetal movement. Conclusion The development of reliable 3D extended-field-of-view ultrasound may help ultrasound practitioners to demonstrate the anatomical relation of pathology and provide a convenient way to store data.
    The British journal of radiology 10/2012; 85(1018):e919-24. · 2.11 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: There is growing evidence that focal thinning of cortical bone in the proximal femur may predispose a hip to fracture. Detecting such defects in clinical CT is challenging, since cortices may be significantly thinner than the imaging system's point spread function. We recently proposed a model-fitting technique to measure sub-millimetre cortices, an ill-posed problem which was regularized by assuming a specific, fixed value for the cortical density. In this paper, we develop the work further by proposing and evaluating a more rigorous method for estimating the constant cortical density, and extend the paradigm to encompass the mapping of cortical mass (mineral mg/cm(2)) in addition to thickness. Density, thickness and mass estimates are evaluated on sixteen cadaveric femurs, with high resolution measurements from a micro-CT scanner providing the gold standard. The results demonstrate robust, accurate measurement of peak cortical density and cortical mass. Cortical thickness errors are confined to regions of thin cortex and are bounded by the extent to which the local density deviates from the peak, averaging 20% for 0.5mm cortex.
    Medical image analysis 02/2012; 16(5):952-65. · 3.09 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Individuals with osteoporosis are predisposed to hip fracture during trips, stumbles or falls, but half of all hip fractures occur in those without generalised osteoporosis. By analysing ordinary clinical CT scans using a novel cortical thickness mapping technique, we discovered patches of markedly thinner bone at fracture-prone regions in the femurs of women with acute hip fracture compared with controls. We analysed CT scans from 75 female volunteers with acute fracture and 75 age- and sex-matched controls. We classified the fracture location as femoral neck or trochanteric before creating bone thickness maps of the outer 'cortical' shell of the intact contra-lateral hip. After registration of each bone to an average femur shape and statistical parametric mapping, we were able to visualise and quantify statistically significant foci of thinner cortical bone associated with each fracture type, assuming good symmetry of bone structure between the intact and fractured hip. The technique allowed us to pinpoint systematic differences and display the results on a 3D average femur shape model. The cortex was generally thinner in femoral neck fracture cases than controls. More striking were several discrete patches of statistically significant thinner bone of up to 30%, which coincided with common sites of fracture initiation (femoral neck or trochanteric). Femoral neck fracture patients had a thumbnail-sized patch of focal osteoporosis at the upper head-neck junction. This region coincided with a weak part of the femur, prone to both spontaneous 'tensile' fractures of the femoral neck, and as a site of crack initiation when falling sideways. Current hip fracture prevention strategies are based on case finding: they involve clinical risk factor estimation to determine the need for single-plane bone density measurement within a standard region of interest (ROI) of the femoral neck. The precise sites of focal osteoporosis that we have identified are overlooked by current 2D bone densitometry methods.
    PLoS ONE 01/2012; 7(6):e38466. · 3.53 Impact Factor
  • Source
    Ultrasound in Obstetrics and Gynecology 10/2011; 38(S1). · 3.56 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Ultrasound elastography is a technique used for clinical imaging of tissue stiffness with a conventional ultrasound machine. It was first proposed two decades ago, but active research continues in this area to the present day. Numerous clinical applications have been investigated, mostly related to cancer imaging, and though these have yet to prove conclusive, the technique has seen increasing commercial and clinical interest. This paper presents a review of the most widely adopted, non-quantitative, techniques focusing on technical innovations rather than clinical applications. The review is not intended to be exhaustive, concentrating instead on placing the various techniques in context according to the authors' perspective of the field.
    Interface focus: a theme supplement of Journal of the Royal Society interface 08/2011; 1(4):540-52. · 2.21 Impact Factor
  • Source
    Ultrasound in Obstetrics and Gynecology 05/2011; 38(2):235-6. · 3.56 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This article presents a new method for acquiring three-dimensional (3-D) volumes of ultrasonic axial strain data. The method uses a mechanically-swept probe to sweep out a single volume while applying a continuously varying axial compression. Acquisition of a volume takes 15-20 s. A strain volume is then calculated by comparing frame pairs throughout the sequence. The method uses strain quality estimates to automatically pick out high quality frame pairs, and so does not require careful control of the axial compression. In a series of in vitro and in vivo experiments, we quantify the image quality of the new method and also assess its ease of use. Results are compared with those for the current best alternative, which calculates strain between two complete volumes. The volume pair approach can produce high quality data, but skillful scanning is required to acquire two volumes with appropriate relative strain. In the new method, the automatic quality-weighted selection of image pairs overcomes this difficulty and the method produces superior quality images with a relatively relaxed scanning technique.
    Ultrasound in medicine & biology 03/2011; 37(3):434-41. · 2.46 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We have recently developed image processing techniques for measuring the cortical thicknesses of skeletal structures in vivo, with resolution surpassing that of the underlying computed tomography system. The resulting thickness maps can be analysed across cohorts by statistical parametric mapping. Applying these methods to the proximal femurs of osteoporotic women, we discover targeted and apparently synergistic effects of pharmaceutical osteoporosis therapy and habitual mechanical load in enhancing bone thickness.
    PLoS ONE 01/2011; 6(1):e16190. · 3.53 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: There are sometimes occasions when ultrasound beamforming is performed with only a subset of the total data that will eventually be available. The most obvious example is a mechanically-swept (wobbler) probe in which the three-dimensional data block is formed from a set of individual B-scans. In these circumstances, non-blind deconvolution can be used to improve the resolution of the data. Unfortunately, most of these situations involve large blocks of three-dimensional data. Furthermore, the ultrasound blur function varies spatially with distance from the transducer. These two facts make the deconvolution process time-consuming to implement. This paper is about ways to address this problem and produce spatially-varying deconvolution of large blocks of three-dimensional data in a matter of seconds. We present two approaches, one based on hardware and the other based on software. We compare the time they each take to achieve similar results and discuss the computational resources and form of blur model that each requires.
    IEEE transactions on ultrasonics, ferroelectrics, and frequency control 01/2011; 58(1):234-8. · 1.80 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: As three-dimensional (3D) ultrasound is becoming more and more popular, there has been increased interest in using a position sensor to track the trajectory of the D ultrasound probe during the scan. One application is the improvement of image quality by fusion of multiple scans from different orientations. With a position sensor mounted on the probe, the clinicians face additional difficulties, for example, maintaining a line-of-sight between the sensor and the reference point. Therefore, the objective of this paper is to register the volumes using an automatic image-based registration technique. In this paper, we employ the particle swarm optimization (PSO) technique to calculate the six rigid-body transformation parameters (three for translation and three for rotation) between successive volumes of D ultrasound data. We obtain vertical and horizontal slices through the acquired volumes and then use an intensity-based similarity measure as a fitness function for each particle. We considered various settings in the PSO to find a set of parameters to give the best convergence. We found the visually acceptable registration when the initial orientations of the particles were confined to within a few degrees of the orientations obtained from position sensor.
    Acoustical Imaging. 01/2011;
  • [Show abstract] [Hide abstract]
    ABSTRACT: In medical pulse-echo ultrasound imaging, a constant sound speed of 1,540 m/s in soft tissues is assumed. When the actual speed is different the mismatch can lead to image distortions. Even if the assumed speed is correct, ultrasound images can be difficult to interpret due to image blurring and the presence of speckle. However, this can be improved by non-blind deconvolution if the point-spread function (PSF) is known. In clinical applications a sufficiently accurate estimate of the PSF is difficult to obtain because of the unknown properties (including speed of sound) of soft tissues. In this paper, we address two topics: first, we explore the sensitivity of our deconvolution algorithm to variations in the speed of sound in the tissue; second, we extend our deconvolution algorithm to enable it to adapt to (and estimate) an unknown sound speed. In the first topic, the results reveal that the deconvolution output is sufficiently sensitive to the accuracy of the sound speed that the speed itself can be estimated using deconvolution. However, qualitative assessment suggests that we may not need the exact speed of sound for successful deconvolution so long as the assumed speed does not deviate significantly from the true value. In the second topic, the goal is gradually to adapt the assumed sound speed to improve the deconvolution and eventually estimate the true sound speed. We tested our algorithm with in vitro phantoms where the estimation error was found to be +0.01 ± 0.60% (mean ± standard deviation). In addition to the speed estimation itself, our method has also proved capable of producing better restoration of the ultrasound images than deconvolution by an assumed speed of 1,540 m/s when this assumption is significantly in error. KeywordsMedical ultrasound image-Non-blind deconvolution-Point-spread function-Complex wavelet transform-Speed of sound-Sound speed estimation
    12/2010: pages 371-378;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Most of the conventional ultrasound machines in hospitals work in two dimensions. However, there are some applications where doctors would like to be able to gather ultrasound data as a three-dimensional (3D) block rather than a two-dimensional (2D) slice. Two different types of 3D ultrasound have been developed to meet this requirement. One type involves a special probe that can record a fixed block of data, either by having an internal sweeping mechanism or by using electronic steering. The other type of 3D ultrasound uses a conventional 2D ultrasound probe together with a position sensor and is called freehand 3D ultrasound. A natural progression of the mechanically-swept 3D ultrasound system is to combine it with the free hand sensor. This results in an extended field of view. There are two major problems with using a position sensor. Firstly, line-of-sight needs to be maintained between the sensor and the reference point. Secondly, the multiple volumes rarely register because of tissue displacement and deformation. Therefore, the objective of this paper is to get rid of the inconvenient position sensor and to use an automatic image-based registration technique. We provide an experimental study of several intensity-based similarity measures for the registration of 3D ultrasound volumes. Rather than choosing a conventional voxel array to represent the 3D blocks, we use corresponding vertical and horizontal image slices from the blocks to be matched. This limits the amount of data thus making the calculation of the similarity measure less computationally expensive. KeywordsSimilarity measures-3D ultrasound-Automatic registration
    12/2010: pages 315-323;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Deformation estimation by block matching plays a central role in most ultrasonic elastography systems. However, not all matches are equally reliable, since the data in many blocks may be severely decorrelated for a number of reasons, including electrical noise, intra-window strain and out-of-plane motion. Conventionally, an estimate of the match quality is derived from the post-alignment similarity metric and used to reject poor matches. This paper moves beyond such rejection thresholds in two ways. First, a displacement tracking strategy is described, in which the tracking direction is not fixed in advance but is generated dynamically according to the quality metric. Second, a nonparametric regression technique is used to smooth the resulting strain images, with more smoothing in low quality regions and less in high quality regions. Simulation and in vivo results show how these two innovations help to improve the accuracy and intelligibility of the strain images. KeywordsStrain imaging- Elastography -Displacement-Deformation
    12/2010: pages 87-95;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The axial-shear strain distribution of soft tissue under load contains information useful for differentiating benign and malignant tumors. This paper describes a novel axial-shear strain normalization method. The algorithm builds on an existing normalization procedure for axial strain to map the shear strain values to the range [ -π/2, π/2]. The normalized shear data do not change sign with the direction of axial probe motion, and therefore can be time averaged without loss of information. Experiments in simulation, in vitro, and in vivo confirm the advantages of normalization. The proposed method is well suited to freehand strain imaging and enables the visualization of subtle slip patterns around inclusions.
    IEEE transactions on ultrasonics, ferroelectrics, and frequency control 12/2010; 57(12):2833-8. · 1.80 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper considers registration of 3D ultrasound volumes acquired in multiple views for display in a single image volume. One way to acquire 3D data is to use a mechanically swept 3D probe. However, the usefulness of these probes is restricted by their limited field of view. This problem can be overcome by attaching a six-degree-of-freedom (DOF) position sensor to the probe, and displaying the information from multiple sweeps in their proper positions. However, an external six-DOF position sensor can be an inconvenience in a clinical setting. The objective of this paper is to propose a hybrid strategy that replaces the sensor with a combination of three-DOF image registration and an unobtrusive inertial sensor for measuring orientation. We examine a range of optimization algorithms and similarity measures for registration and compare them in in vitro and in vivo experiments. We register based on multiple reslice images rather than a whole voxel array. In this paper, we use a large number of reslices for improved reliability at the expense of computational speed. We have found that the Levenberg–Marquardt method is very fast but is not guaranteed to give the correct solution all the time. We conclude that normalized mutual information used in the Nelder–Mead simplex algorithm is potentially suitable for the registration task with an average execution time of around 5 min, in the majority of cases, with two restarts in a C++ implementation on a 3.0 GHz Intel Core 2 Duo CPU machine.
    Measurement Science and Technology 07/2010; 21(8):085803. · 1.44 Impact Factor

Publication Stats

2k Citations
209.72 Total Impact Points

Institutions

  • 1991–2014
    • University of Cambridge
      • • Department of Engineering
      • • Department of Radiology
      Cambridge, England, United Kingdom
  • 2013
    • University of Glasgow
      • School of Engineering
      Glasgow, SCT, United Kingdom
  • 2000
    • CSU Mentor
      Long Beach, California, United States