Showing papers by "Simon R. Arridge published in 2008"

Gauss–Newton reconstruction method for optical tomography using the finite element solution of the radiative transfer equation

Abstract: The radiative transfer equation can be utilized in optical tomography in situations in which the more commonly applied diffusion approximation is not valid. In this paper, an image reconstruction method based on a frequency domain radiative transfer equation is developed. The approach is based on a total variation output regularized least squares method which is solved with a Gauss–Newton algorithm. The radiative transfer equation is numerically solved with a finite element method in which both the spatial and angular discretizations are implemented in piecewise linear bases. Furthermore, the streamline diffusion modification is utilized to improve the numerical stability. The approach is tested with simulations. Reconstructions from different cases including domains with low-scattering regions are shown. The results show that the radiative transfer equation can be utilized in optical tomography and it can produce good quality images even in the presence of low-scattering regions.

Model-Based Imaging of Cardiac Apparent Conductivity and Local Conduction Velocity for Diagnosis and Planning of Therapy

Abstract: We present an adaptive algorithm which uses a fast electrophysiological (EP) model to estimate apparent electrical conductivity and local conduction velocity from noncontact mapping of the endocardial surface potential. Development of such functional imaging revealing hidden parameters of the heart can be instrumental for improved diagnosis and planning of therapy for cardiac arrhythmia and heart failure, for example during procedures such as radio-frequency ablation and cardiac resynchronisation therapy. The proposed model is validated on synthetic data and applied to clinical data derived using hybrid X-ray/magnetic resonance imaging. We demonstrate a qualitative match between the estimated conductivity parameter and pathology locations in the human left ventricle. We also present a proof of concept for an electrophysiological model which utilizes the estimated apparent conductivity parameter to simulate the effect of pacing different ventricular sites. This approach opens up possibilities to directly integrate modelling in the cardiac EP laboratory.

An Atlas-Based Segmentation Propagation Framework Using Locally Affine Registration --- Application to Automatic Whole Heart Segmentation

Abstract: In this paper, we present a novel registration algorithm for locally affine registrations. This method preserves the anatomical and intensity class relationships between the local regions. A regularisation procedure is used to maintain a global diffeomorphic transformation. Combined with a novel generic method for accurately inverting the final deformation field, we include our techniques within an atlas-based segmentation propagation framework. We applied our method to automatically segment the whole heart from cardiac magnetic resonance images from a cohort of 18 volunteers (acquisition resolution 2 × 2 × 2 mm). The results show that the proposed method provides a robust initialisation for the atlas-based segmentation propagation framework refined with a fluid registration. We validated our approach against other registration strategies, and demonstrated that we improved the accuracy of the whole heart segmentations (1.8 ± 0.42 mm).

Model-based imaging of cardiac apparent conductivity and local conduction velocity for diagnosis and planning of therapy. IEEE Trans Med Imaging.

Abstract: We present an adaptive algorithm which uses a fast electrophysiological (EP) model to estimate apparent electrical conductivity and local conduction velocity from noncontact mapping of the endocardial surface potential. Development of such functional imaging revealing hidden parameters of the heart can be instrumental for improved diagnosis and planning of therapy for cardiac arrhythmia and heart failure, for example during procedures such as radio-frequency ablation and cardiac resynchronisation therapy. The proposed model is validated on synthetic data and applied to clinical data derived using hybrid X-ray/magnetic resonance imaging. We demonstrate a qualitative match between the estimated conductivity parameter and pathology locations in the human left ventricle. We also present a proof of concept for an electrophysiological model which utilizes the estimated apparent conductivity parameter to simulate the effect of pacing different ventricular sites. This approach opens up possibilities to directly integrate modelling in the cardiac EP laboratory.

Temporal propagation of spatial information in turbid media

Abstract: The time-resolved propagation of spatial frequencies in turbid media is investigated. We discuss the relevant information content in terms of the spatial Fourier component of the propagating pulse. We present a method to record and analyze the time-gated spatial frequency response of highly scattering media. We demonstrate that high spatial frequency information appears in the early time-gated signal, whereas low frequencies persist for longer times.

Comparison of methods for optimal choice of the regularization parameter for linear electrical impedance tomography of brain function

Abstract: Electrical impedance tomography has the potential to provide a portable non-invasive method for imaging brain function. Clinical data collection has largely been undertaken with time difference data and linear image reconstruction methods. The purpose of this work was to determine the best method for selecting the regularization parameter of the inverse procedure, using the specific application of evoked brain activity in neonatal babies as an exemplar. The solution error norm and image SNR for the L-curve (LC), discrepancy principle (DP), generalized cross validation (GCV) and unbiased predictive risk estimator (UPRE) selection methods were evaluated in simulated data using an anatomically accurate finite element method (FEM) of the neonatal head and impedance changes due to blood flow in the visual cortex recorded in vivo. For simulated data, LC, GCV and UPRE were equally best. In human data in four neonatal infants, no significant differences were found among selection methods. We recommend that GCV or LC be employed for reconstruction of human neonatal images, as UPRE requires an empirical estimate of the noise variance.

Parameter and structure reconstruction in optical tomography

Abstract: Many reconstruction problems in optical tomography, such as the imaging of haematoma in the brain, or breast tumour screening, require the detection and localisation of well-defined objects on a homogeneous or weakly varying background. A conventional parameter identification approach is to represent the image in a finite dimensional set of basis functions and to find the coefficients of this set using an optimisation strategy. Shape-based reconstruction techniques on the other hand, seek to find the boundaries of objects and possibly a representation of the interior either as constant or slowly varying functions. In this paper we compare both explicit and implicit shape reconstruction methods for the simultaneous recovery of absorption and diffusion inclusions in a three-dimensional scattering medium. Images reconstructed from simulated frequency-domain boundary measurements are compared to a voxel-based conjugate gradient method. The results demonstrate the feasibility of the shape based methods.

3D Digital Breast Tomosynthesis Using Total Variation Regularization

Abstract: 3D digital breast imaging promises to significantly reduce both false negatives and false positives, allowing the earlier detection of cancer while reducing the occurrence of callbacks. In this paper we have developed an iterative algorithm that improves on the quality of the reconstructed images over current algorithms like filtered back projection. Our algorithm uses Total Variation to restrain the noise while enhancing edge features. We have applied this algorithm to data obtained using the XCounter mammography system.

Utilizing the radiative transfer equation in optical tomography

Abstract: We propose a method which utilizes the radiative transfer equation in optical tomography. In this approach, the radiative transfer equation is used as light propagation model in those regions in which the assumptions of the diffusion theory are not valid and the diffusion approximation is used elsewhere. Both the radiative transfer equation and the diffusion approximation are numerically solved with a finite element method. In the finite element solution of the radiative transfer equation, both the spatial and angular discretizations are implemented in piecewise linear bases.

3-D shape and contrast reconstruction in optical tomography with level sets

Abstract: Many applications of optical tomography in medical diagnostics, including the imaging of haematoma and tumours or the localisation of organs marked by a contrast agent, require the detection and localisation of well-defined boundaries between a homogeneous or weakly varying background and inclusions of different optical parameters. Shape-based reconstruction techniques, such as level sets, are better suited to solve these problems than conventional voxel-based approaches, which often lead to blurring of the boundaries of features and loss of contrast. In this paper we present a level set technique for the simultaneous recovery of absorption and diffusion distributions in a three-dimensional scattering medium. Images reconstructed from simulated frequency-domain boundary measurements are compared to a voxel-based conjugate gradient method. The results demonstrate the feasibility of the level set method.

Adjoint time domain method for fluorescent imaging in turbid media

Abstract: Application of adjoint time domain methods to the inverse problem in 3D fluorescence imaging is a novel approach. We demonstrate the feasibility of this approach experimentally on the basis of a time gating technique completely in the time domain by using a small number of time windows. The evolution of the fluorescence energy density function inside a highly scattering cylinder was reconstructed together with optical parameters. Reconstructed energy density was used in localizing two fluorescent tubes. Relatively accurate reconstruction demonstrates the effectiveness and the potential of the proposed technique.

Simultaneous estimation of chromophore concentration and scattering distributions from multiwavelength photoacoustic images

Abstract: In biomedical photoacoustic tomography of soft tissue, the initial acoustic pressure distribution following the absorption of a short excitation laser pulse, is recovered as a function of position. This initial pressure distribution is proportional to the absorbed optical energy density, and is thus related (albeit indirectly) to the tissue optical coefficients. When imaging soft tissue which contains several absorbing chromophores (such as oxy- and deoxy-haemoglobin, water, etc.), the primary quantity of interest is the concentrations of the chromophores at each point in the tissue, and not the absorbed optical energy density, which is nonlinearly related to the chromophore concentrations, and also depends on the distribution of scattering. Estimating the distribution of the concentration of a chromophore therefore requires the recovery of two unknown functions (chromophore concentration and scattering distributions) from measurements of one (absorbed energy density). For measurements made at a single optical wavelength, this problem suffers from nonuniqueness, and cannot be solved without additional information being incorporated. A simulated example is used here to demonstrate that, in principle, by using multi-wavelength data and incorporating the known wavelength dependence of the chromophore absorption and the scattering as prior information, a chromophore concentration and spatial dependence of the scattering can be recovered simultaneously. This step opens the way to physiological and molecular imaging using multispectral photoacoustic tomography.

Reconstruction of Simple Geometric Objects in 3D Optical Tomography Using an Adjoint Technique and a Boundary Element Method

Abstract: In this paper we consider the recovery of ellipsoidal 3D shapes with piecewise constant coefficients in Diffuse Optical Tomography (DOT). We use an adjoint scheme for calculating gradients for the shape parameters defining the unknown ellipsoids, and a Newton-type optimisation process for the minimization of a least squares data misfit functional. A boundary integral formulation is used for the forward modelling. An advantage of the proposed method is the implicit regularisation effect arising from the reduced dimensionality of the inverse problem. Results of a numerical experiment in 3D are shown which demonstrate the performance of the method.

Photoacoustic tomography using reverberant field data from a single detector

Abstract: In biomedical photoacoustic tomography (PAT), ultrasonic pulses generated by the absorption of near-infrared light are recorded over an array of detectors, and the measured pressure time series are used to recover an image of the initial acoustic pressure distribution within the tissue, which is related to the tissue optical coefficients and therefore to tissue physiology. For high resolution imaging, large-area detector arrays with a high density of sensitive, small elements are required. Such arrays can be expensive, so reverberant-field PAT has been suggested as a means of obtaining PAT images using arrays with a smaller number of detectors or even a single detector. We propose that by recording the reflections from a reverberant cavity in addition to the primary acoustic waves, sufficient information can be captured to allow a PAT image to be reconstructed, without the requirement for a large-area array. A pilot study using simple 2D simulations, backprojections and modal inversions was undertaken to assess the feasibility of this approach to PAT.

Sensitivity and specificity of 3d optical mammography

Abstract: Optical tomography is being developed to detect and specify disease in the female breast.Assessors were trained to interpret optical images, then presented with images from further patients.The sensitivity was 85.8% and the specificity 66.8%.

3D optical mammography of the uncompressed breast

Abstract: We have successfully performed optical tomography of the uncompressed breast on 52 volunteers with breast lesions. We describe a new method for image reconstruction which draws prior information from the optical image itself.

Time-Domain Fluorescence Lifetime Tomography

Abstract: We present a platform for fluorescence lifetime tomography utilising tuneable supercontinuum excitation and wide-field time-gated technology. Applied to optical projection and diffuse fluorescence tomography, we demonstrate 3-D time-resolved fluorescence reconstruction in transparent and scattering phantoms.

Multispectral Fluorescence Enhanced Diffuse Optical Tomography Evaluated with Weight Matrix Free Algorithm

Abstract: We present a novel multispectral scheme for fluorescence enhanced diffuse optical tomography. Reconstructions are performed using a weight matrix free algorithm. Initial multispectral reconstructions are shown.

3 D fluorescence imaging in turbid media by using time gated data acquisition

Abstract: We demonstrate the feasibility of fluorescence imaging experimentally on the basis of a time gating technique completely in the time domain by using a small number of time steps.