-
[show abstract]
[hide abstract]
ABSTRACT: The problem of respiratory motion has proved a serious obstacle in developing techniques to acquire images or guide interventions in abdominal and thoracic organs. Motion models offer a possible solution to these problems, and as a result the field of respiratory motion modelling has become an active one over the past 15years. A motion model can be defined as a process that takes some surrogate data as input and produces a motion estimate as output. Many techniques have been proposed in the literature, differing in the data used to form the models, the type of model employed, how this model is computed, the type of surrogate data used as input to the model in order to make motion estimates and what form this output should take. In addition, a wide range of different application areas have been proposed. In this paper we summarise the state of the art in this important field and in the process highlight the key papers that have driven its advance. The intention is that this will serve as a timely review and comparison of the different techniques proposed to date and as a basis to inform future research in this area.
Medical image analysis 10/2012; · 3.09 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Organ and tumour motion has a significant impact on the planning and delivery of radiotherapy treatment. At present imaging modality such as four-dimensional computer tomography (4DCT) cannot be used to measure the variability of motion between different respiratory cycles. To create reliable motion models, one needs to acquire volumetric data sets of the lungs with sufficient sampling of the breathing cycle. In this paper we investigate the use of highly parallel MRI to acquire such data. A 32 channel coil in conjunction with a balanced SSFP sequence and a SENSE factor of 6 were used to acquire volumetric data sets in five healthy volunteers. The acquisition was repeated for seven series of different breathing patterns. The data acquired was of sufficient spatial resolution (5 × 5 × 5 mm(3)) and image quality to carry out automated non-rigid registration. The acquisition rate (c.a. 2 volumes per second) allowed for a meaningful sampling of the different respiratory curves that were automatically obtained from the skin surface motion. This acquisition technique should provide images of high enough quality to create statistical respiratory models.
Physica Medica 03/2012; · 1.07 Impact Factor
-
Physics in Medicine and Biology 01/2012; · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: A dynamic-contrast-enhanced magnetic resonance imaging (DCE-MRI) dataset consists of many imaging frames, often acquired both before and after contrast injection. Due to the length of time spent acquiring images, patient motion is likely and image re-alignment or registration is required before further analysis such as pharmacokinetic model fitting. Non-rigid image registration procedures may be used to correct motion artefacts; however, a careful choice of registration strategy is required to reduce misregistration artefacts associated with enhancing features. This work investigates the effect of registration on the results of model-fitting algorithms for 52 DCE-MR mammography cases for 14 patients. Results are divided into two sections: a comparison of registration strategies in which a DCE-MRI-specific algorithm is preferred in 50% of cases, followed by an investigation of parameter changes with known applied deformations, inspecting the effect of magnitude and timing of motion artefacts. Increased motion magnitude correlates with increased model-fit residual and is seen to have a strong influence on the visibility of strongly enhancing features. Motion artefacts in images close to the contrast agent arrival have a disproportionate effect on discrepancies in parameter estimation. The choice of algorithm, magnitude of motion and timing of the motion are each shown to influence estimated pharmacokinetic parameters even when motion magnitude is small.
Physics in Medicine and Biology 11/2011; 56(24):7693-708. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The registration of temporal 2D X-ray mammograms is a challenging task which is difficult in part due to the superimposition of structures found in the breast together with the complex 3D deformation that the breast undergoes during compression. This paper presents an investigation into a method for registration of 2D mammographic images which accounts for varying displacements through the breast by allowing additional degrees of freedom for the transformation at different depths. We use simulated compressions of MR derived volumes to determine what is the maximum sampling interval of the deformation field that yields a deformation field that agrees with the simulation within a certain tolerance. We found that a subsampling of the deformation field by a factor of 17 in-plane and 10 out-of-plane increased error in the estimate of displacement of tissue projected onto the imaging plane by less than 1mm for all points, excluding outliers. This has important ramification for the search for the appropriate 2D non-diffeomorphic transformation between 2 2D X-ray mammograms. This work supports the development of more accurate registration algorithms by taking into account the realistic 3D movement of tissue as well as reducing the complexity of the problem.
Biomedical Imaging: From Nano to Macro, 2011 IEEE International Symposium on; 05/2011
-
[show abstract]
[hide abstract]
ABSTRACT: Image registration algorithms are widely used in medical imaging to spatially align anatomical features. This work investigates the regularisation of the registration transformation by unification of the standard diffusion equation regulariser with the standard curvature regulariser. A variational nonrigid registration scheme is employed with periodic boundary conditions and tested over a range of regularisation parameters. A variable regularisation algorithm is also proposed that automatically adapts the regularisation as the registration progresses. Improved registration performance on two simulated biomechanical deformations of 3D breast MRI over four subjects is observed; correcting the simulated deformation residual by 69% and 73% respectively for diffusion regularisation, 36% and 45% for curvature regularisation and up to 73% and 75% for variable regularisation.
Biomedical Imaging: From Nano to Macro, 2011 IEEE International Symposium on; 05/2011
-
[show abstract]
[hide abstract]
ABSTRACT: Respiratory motion can vary dramatically between the planning stage and the different fractions of radiotherapy treatment. Motion predictions used when constructing the radiotherapy plan may be unsuitable for later fractions of treatment. This paper presents a methodology for constructing patient-specific respiratory motion models and uses these models to evaluate and analyse the inter-fraction variations in the respiratory motion. The internal respiratory motion is determined from the deformable registration of Cine CT data and related to a respiratory surrogate signal derived from 3D skin surface data. Three different models for relating the internal motion to the surrogate signal have been investigated in this work. Data were acquired from six lung cancer patients. Two full datasets were acquired for each patient, one before the course of radiotherapy treatment and one at the end (approximately 6 weeks later). Separate models were built for each dataset. All models could accurately predict the respiratory motion in the same dataset, but had large errors when predicting the motion in the other dataset. Analysis of the inter-fraction variations revealed that most variations were spatially varying base-line shifts, but changes to the anatomy and the motion trajectories were also observed.
Physics in Medicine and Biology 01/2011; 56(1):251-72. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Magnetic resonance (MR) imaging has become a routine modality for the determination of patient cardiac morphology. The extraction of this information can be important for the development of new clinical applications as well as the planning and guidance of cardiac interventional procedures. To avoid inter- and intra-observer variability of manual delineation, it is highly desirable to develop an automatic technique for whole heart segmentation of cardiac magnetic resonance images. However, automating this process is complicated by the limited quality of acquired images and large shape variation of the heart between subjects. In this paper, we propose a fully automatic whole heart segmentation framework based on two new image registration algorithms: the locally affine registration method (LARM) and the free-form deformations with adaptive control point status (ACPS FFDs). LARM provides the correspondence of anatomical substructures such as the four chambers and great vessels of the heart, while the registration using ACPS FFDs refines the local details using a constrained optimization scheme. We validated our proposed segmentation framework on 37 cardiac MR volumes on the end-diastolic phase, displaying a wide diversity of morphology and pathology, and achieved a mean accuracy of 2.14 ± 0.63 mm (rms surface distance) and a maximal error of 4.31 mm.
IEEE Transactions on Medical Imaging 10/2010; · 3.64 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Any similarity measure used for image registration depends in some way on the region Ω describing the overlap between the floating and reference images. In variational registration, where the Gâteaux derivative of the similarity measure drives the registration, most literature implicitly assumes that Ω remains constant. This assumption is valid if homogeneous Dirichlet or sliding boundary conditions are chosen for the displacement field; however, it is invalid if any other type of boundary conditions are chosen, or if the similarity measure is computed over some masked portion of the overlap region. This article illustrates how these more general situations of different boundary conditions and/or masked regions can be accommodated in variational registration by explicitly accounting for the varying Ω in the Gâteaux derivative of the similarity measure.
Biomedical Imaging: From Nano to Macro, 2010 IEEE International Symposium on; 05/2010
-
[show abstract]
[hide abstract]
ABSTRACT: The acquisition of Diffusion Weighted MR images may be confounded by both patient motion and machine eddy currents, these may potentially be corrected by image registration. Two non-rigid registration schemes are compared to the result of an affine registration: a single fluid registration of the individual diffusion directions and a Progressive Principal Component Registration. All registrations are full 3D. 12 DW-MRI datasets consisting of 128×128×64 volumes from 15 diffusion directions are registered by each method and the results combined to produce fractional anisotropy maps. These maps are then inspected for improved feature appearance and fractional anisotropy variability. The affine registration demonstrates a modest improvement; image alignment by single fluid registration causes lateral brain features to appear sharper at the expense of poor deformations of the medial brain; registration by PPCR demonstrates improved contrast of lateral brain features and lower fractional anisotropy variability.
Biomedical Imaging: From Nano to Macro, 2010 IEEE International Symposium on; 05/2010
-
[show abstract]
[hide abstract]
ABSTRACT: Intensity inhomogeneity is an important phenomenon affecting magnetic resonance imaging, which can greatly affect computational image analysis. Bias correction algorithms are commonly used, but are imperfect, leaving residual inhomogeneity that will usually differ in serial images of the same subject. This differential bias can have a detrimental effect on further processing, such as registration and quantification of small longitudinal changes. We embed a differential bias field model within a nonrigid registration framework. The spatial transformation and DBC parameters are optimised concurrently using the normalised mutual information as a metric. We show significant reductions in registration error with the proposed framework.
Biomedical Imaging: From Nano to Macro, 2010 IEEE International Symposium on; 05/2010
-
[show abstract]
[hide abstract]
ABSTRACT: As the number of clinical applications requiring nonrigid image registration continues to grow, it is important to design registration algorithms that not only build on the best available theory, but also are computationally efficient. Thirion's Demons algorithm [1] estimates nonrigid deformations by successively estimating force vectors that drive the deformation toward alignment, and then smoothing the force vectors by convolution with a Gaussian kernel. It essentially approximates a deformation under diffusion regularization [2], and it is a popular choice of algorithm for nonrigid registration because of its linear computational complexity and ease of implementation. In this article, we show how the Demons algorithm can be generalized to handle other common regularizers, yielding O(n) algorithms that employ Gaussian convolution for elastic, fluid, and curvature registration. We compare the speed of the proposed algorithms with algorithms based on Fourier methods [3] for registering serial chest CT studies.
Biomedical Imaging: From Nano to Macro, 2009. ISBI '09. IEEE International Symposium on; 08/2009
-
[show abstract]
[hide abstract]
ABSTRACT: This article presents a method to reconstruct liver MRI data acquired continuously during free breathing, without any external sensor or navigator measurements. When the deformations associated with k-space data are known, generalized matrix inversion reconstruction has been shown to be effective in reducing the ghosting and blurring artifacts of motion. This article describes a novel method to obtain these nonrigid deformations. A breathing model is built from a fast dynamic series: low spatial resolution images are registered and their deformations parameterized by overall superior-inferior displacement. The correct deformation for each subset of the subsequent imaging data is then found by comparing a few lines of k-space with the equivalent lines from a deformed reference image while varying the deformation over the model parameter. This procedure is known as image deformation recovery using overlapping partial samples (iDROPS). Simulations using 10 rapid dynamic studies from volunteers showed the average error in iDROPS-derived deformations within the liver to be 1.43 mm. A further four volunteers were imaged at higher spatial resolution. The complete reconstruction process using data from throughout several breathing cycles was shown to reduce blurring and ghosting in the liver. Retrospective respiratory gating was also demonstrated using the iDROPS parameterization.
Magnetic Resonance in Medicine 06/2009; 62(2):440-9. · 2.96 Impact Factor
-
A P King,
R Boubertakh,
K S Rhode,
Y L Ma,
P Chinchapatnam,
G Gao,
T Tangcharoen,
M Ginks,
M Cooklin,
J S Gill, D J Hawkes,
R S Razavi,
T Schaeffter
[show abstract]
[hide abstract]
ABSTRACT: We describe a system for respiratory motion correction of MRI-derived roadmaps for use in X-ray guided cardiac catheterisation procedures. The technique uses a subject-specific affine motion model that is quickly constructed from a short pre-procedure MRI scan. We test a dynamic MRI sequence that acquires a small number of high resolution slices, rather than a single low resolution volume. Additionally, we use prior knowledge of the nature of cardiac respiratory motion by constraining the model to use only the dominant modes of motion. During the procedure the motion of the diaphragm is tracked in X-ray fluoroscopy images, allowing the roadmap to be updated using the motion model. X-ray image acquisition is cardiac gated. Validation is performed on four volunteer datasets and three patient datasets. The accuracy of the model in 3D was within 5mm in 97.6% of volunteer validations. For the patients, 2D accuracy was improved from 5 to 13mm before applying the model to 2-4mm afterwards. For the dynamic MRI sequence comparison, the highest errors were found when using the low resolution volume sequence with an unconstrained model.
Medical image analysis 02/2009; 13(3):419-31. · 3.09 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Efficient and accurate techniques for simulation of soft tissue deformation are an increasingly valuable tool in many areas of medical image computing, such as biomechanically-driven image registration and interactive surgical simulation. For reasons of efficiency most analyses are based on simplified linear formulations, and previously almost all have ignored well established features of tissue mechanical response such as anisotropy and time-dependence. We address these latter issues by firstly presenting a generalised anisotropic viscoelastic constitutive framework for soft tissues, particular cases of which have previously been used to model a wide range of tissues. We then develop an efficient solution procedure for the accompanying viscoelastic hereditary integrals which allows use of such models in explicit dynamic finite element algorithms. We show that the procedure allows incorporation of both anisotropy and viscoelasticity for as little as 5.1% additional cost compared with the usual isotropic elastic models. Finally we describe the implementation of a new GPU-based finite element scheme for soft tissue simulation using the CUDA API. Even with the inclusion of more elaborate constitutive models as described the new implementation affords speed improvements compared with our recent graphics API-based implementation, and compared with CPU execution a speed up of 56.3 x is achieved. The validity of the viscoelastic solution procedure and performance of the GPU implementation are demonstrated with a series of numerical examples.
Medical image analysis 11/2008; 13(2):234-44. · 3.09 Impact Factor
-
Medical image analysis 10/2008; 12:586-602. · 3.09 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: This paper describes the construction of 3D statistical deformation models (SDMs) from biomechanical simulations. The method
was used to capture the average breast motion and its variability due to compressing it between two plates as performed for
X-ray mammography. One SDM described the motion from the compressed to the undeformed state. Another SDM captured the deformation
difference due to variations in patient positioning and compression magnitude. Such models could prove useful for guiding
the development of algorithms to register serial X-ray mammograms. The SDMs are based on simulating plausible breast compressions
for a population of 20 patients via finite element models created from segmented 3D MR breast images. Tissue properties and
boundary conditions were varied according to reported values. Three compression configurations (called current, prior1, prior2)
were simulated per breast. The associated displacement fields were mapped into a common space and SDMs were generated using
principle component analysis. Leave-one-patient-out tests showed that these models can reduce the mean error of unseen deformations
on average by 87% (19.35mm to 2.49mm for current-to-undeformed, 13.49mm to 1.70mm for current-to-prior) when using the
first 16 modes of variation.
07/2008: pages 426-432;
-
[show abstract]
[hide abstract]
ABSTRACT: In Studholme et al. introduced normalized mutual information (NMI) as an overlap invariant generalization of mutual information (MI). Even though Studholme showed how NMI could be used effectively in multimodal medical image alignment, the overlap invariance was only established empirically on a few simple examples. In this paper, we illustrate a simple example in which NMI fails to be invariant to changes in overlap size, as do other standard similarity measures including MI, cross correlation (CCorr), correlation coefficient (CCoeff), correlation ratio (CR), and entropy correlation coefficient (ECC). We then derive modified forms of all of these similarity measures that are proven to be invariant to changes in overlap size. This is done by making certain assumptions about background statistics. Experiments on multimodal rigid registration of brain images show that 1) most of the modified similarity measures outperform their standard forms, and 2) the modified version of MI exhibits superior performance over any of the other similarity measures for both CT/MR and PET/MR registration.
Computer Vision and Pattern Recognition Workshops, 2008. CVPRW '08. IEEE Computer Society Conference on; 07/2008
-
Medical Physics 07/2008; 35:3302-3316. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: For assessment of cerebrovascular diseases, it is beneficial to obtain information about the hemodynamics of the vessel system. Recently, we presented a method to quantify blood flow in a single blood vessel segment from rotational angiography. In this paper, we extend the method to bifurcations. A model- based method is proposed which estimates the mean flow, the waveform and the flow division at the bifurcation. The method is validated on experimental data using a phantom for a healthy and a stenosed carotid bifurcation. The average error for the estimate of the flow division was 8%.
Biomedical Imaging: From Nano to Macro, 2008. ISBI 2008. 5th IEEE International Symposium on; 06/2008
