Core Imaging Library - Part I: a versatile Python framework for tomographic imaging.

We present the Core Imaging Library (CIL), an open-source Python framework for tomographic imaging with particular emphasis on reconstruction of challenging datasets. Conventional filtered back-projection reconstruction tends to be insufficient for highly noisy, incomplete, non-standard or multi-channel data arising for example in dynamic, spectral and in situ tomography. CIL provides an extensive modular optimization framework for prototyping reconstruction methods including sparsity and total variation regularization, as well as tools for loading, preprocessing and visualizing tomographic data. The capabilities of CIL are demonstrated on a synchrotron example dataset and three challenging cases spanning golden-ratio neutron tomography, cone-beam X-ray laminography and positron emission tomography. This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 2'.

Employing temporal self-similarity across the entire time domain in computed tomography reconstruction.

There are many cases where one needs to limit the X-ray dose, or the number of projections, or both, for high frame rate (fast) imaging. Normally, it improves temporal resolution but reduces the spatial resolution of the reconstructed data. Fortunately, the redundancy of information in the temporal domain can be employed to improve spatial resolution. In this paper, we propose a novel regularizer for iterative reconstruction of time-lapse computed tomography. The non-local penalty term is driven by the available prior information and employs all available temporal data to improve the spatial resolution of each individual time frame. A high-resolution prior image from the same or a different imaging modality is used to enhance edges which remain stationary throughout the acquisition time while dynamic features tend to be regularized spatially. Effective computational performance together with robust improvement in spatial and temporal resolution makes the proposed method a competitive tool to state-of-the-art techniques.

Comparing D-bar and common regularization-based methods for electrical impedance tomography.

OBJECTIVE: To compare D-bar difference reconstruction with regularized linear reconstruction in electrical impedance tomography. APPROACH: A standard regularized linear approach using a Laplacian penalty and the GREIT method for comparison to the D-bar difference images. Simulated data was generated using a circular phantom with small objects, as well as a 'Pac-Man' shaped conductivity target. An L-curve method was used for parameter selection in both D-bar and the regularized methods. MAIN RESULTS: We found that the D-bar method had a more position independent point spread function, was less sensitive to errors in electrode position and behaved differently with respect to additive noise than the regularized methods. SIGNIFICANCE: The results allow a novel pathway between traditional and D-bar algorithm comparison.

Methods for calculating the electrode position Jacobian for impedance imaging.

Electrical impedance tomography (EIT) or electrical resistivity tomography (ERT) current and measure voltages at the boundary of a domain through electrodes. SIGNIFICANCE: The movement or incorrect placement of electrodes may lead to modelling errors that result in significant reconstructed image artifacts. These errors may be accounted for by allowing for electrode position estimates in the model. Movement may be reconstructed through a first-order approximation, the electrode position Jacobian. A reconstruction that incorporates electrode position estimates and conductivity can significantly reduce image artifacts. Conversely, if electrode position is ignored it can be difficult to distinguish true conductivity changes from reconstruction artifacts which may increase the risk of a flawed interpretation. OBJECTIVE: In this work, we aim to determine the fastest, most accurate approach for estimating the electrode position Jacobian. APPROACH: Four methods of calculating the electrode position Jacobian were evaluated on a homogeneous halfspace. MAIN RESULTS: Results show that Fréchet derivative and rank-one update methods are competitive in computational efficiency but achieve different solutions for certain values of contact impedance and mesh density.

An adaptive current tomography using voltage sources.

This paper describes the development of an adaptive electric current tomography system which contains a novel front-end analog architecture. Programmable voltage sources have been used to deliver currents into the study object and to avoid the difficulties of obtaining high quality current sources. Through inverting an admittance matrix, the system is capable of achieving a desired current drive pattern by applying a computed voltage pattern. The tomograph, operating at 9.6 kHz, comprises 32 driving electrodes and 32 voltage measurement electrodes. The study of system noise performance shows high SNR in the data acquisition which is enhanced by a digital demodulation scheme. In vitro reconstruction images have been obtained with the data collected by the tomograph.

Time series of EIT chest images using singular value decomposition and Fourier transform.

The aim of this study is to propose a useful method for exploring regional ventilation and perfusion in the chest. The paper describes two methods based on singular value decomposition (SVD) and Fourier transform (FT) respectively. This work shows that power spectral density (PSD) and phase images (derived from the Fourier transform) are easier to interpret and more useful tools for exploiting in vivo EIT data in healthy volunteers in order to explore the cardiovascular and respiratory systems.

Generalized optimal current patterns and electrical safety in EIT.

There are a number of constraints which limit the current and voltages which can be applied on a multiple drive electrical imaging system. One obvious constraint is to limit the maximum ohmic power dissipated in the body. Current patterns optimizing distinguishability with respect to this constraint are singular functions of the difference of transconductance matrices with respect to the power norm (the optimal currents of Isaacson). If one constrains the total current (L1 norm) the optimal patterns are pair drives. On the other hand if one constrains the maximum current on each drive electrode (an L(infinity) norm), the optimal patterns have each drive channel set to the maximum source or sink current value. In this paper we consider appropriate safety constraints and discuss how to find the optimal current patterns with those constraints.

Sensitivity matrix and reconstruction algorithm for EIT assuming axial uniformity.

In electrical impedance tomography (EIT) two-dimensional models continue to be applied despite their known inability to provide correct reconstruction. In this paper, a reconstruction algorithm that assumes a translationally invariant conductivity distribution is described. A more precise forward solver is obtained by taking off-slice currents into consideration. An appropriate sensitivity matrix is derived. Numerical evidence for the improvement in precision compared to two-dimensional reconstruction is given.

Development of a real-time adaptive current tomograph.

Following the successful development of a multiple-drive electrical impedance tomography system OXPACT-II featuring a voltage-driven current method for in vitro studies, research work currently being undertaken at the EIT research group in Oxford is aimed at developing a real-time multiple-drive adaptive system, called the Oxford Brookes Adaptive Current Tomograph Mark-III (OXBACT-III) which will operate at several frequencies in between 10-160 kHz. The objective of this system development is to enable EIT clinical studies to be undertaken based on the adaptive current method. One of the most important issues addressed in the new system design is to achieve high data acquisition speed while maintaining sufficient system accuracy. This paper will describe the overall data acquisition system structure and relevant system performance specifications.

A MATLAB package for the EIDORS project to reconstruct two-dimensional EIT images.

The EIDORS (electrical impedance and diffuse optical reconstruction software) project aims to produce a software system for reconstructing images from electrical or diffuse optical data. MATLAB is a software that is used in the EIDORS project for rapid prototyping, graphical user interface construction and image display. We have written a MATLAB package (http://venda.uku.fi/ vauhkon/) which can be used for two-dimensional mesh generation, solving the forward problem and reconstructing and displaying the reconstructed images (resistivity or admittivity). In this paper we briefly describe the mathematical theory on which the codes are based on and also give some examples of the capabilities of the package.

Mutual information as a measure of image quality for 3D dynamic lung imaging with EIT.

We report on a pilot study of dynamic lung electrical impedance tomography (EIT) at the University of Manchester. Low-noise EIT data at 100 frames per second were obtained from healthy male subjects during controlled breathing, followed by magnetic resonance imaging (MRI) subsequently used for spatial validation of the EIT reconstruction. The torso surface in the MR image and electrode positions obtained using MRI fiducial markers informed the construction of a 3D finite element model extruded along the caudal-distal axis of the subject. Small changes in the boundary that occur during respiration were accounted for by incorporating the sensitivity with respect to boundary shape into a robust temporal difference reconstruction algorithm. EIT and MRI images were co-registered using the open source medical imaging software, 3D Slicer. A quantitative comparison of quality of different EIT reconstructions was achieved through calculation of the mutual information with a lung-segmented MR image. EIT reconstructions using a linear shape correction algorithm reduced boundary image artefacts, yielding better contrast of the lungs, and had 10% greater mutual information compared with a standard linear EIT reconstruction.

Acceleration of image-based resolution modelling reconstruction using an expectation maximization nested algorithm.

Recent studies have demonstrated the benefits of a resolution model within iterative reconstruction algorithms in an attempt to account for effects that degrade the spatial resolution of the reconstructed images. However, these algorithms suffer from slower convergence rates, compared to algorithms where no resolution model is used, due to the additional need to solve an image deconvolution problem. In this paper, a recently proposed algorithm, which decouples the tomographic and image deconvolution problems within an image-based expectation maximization (EM) framework, was evaluated. This separation is convenient, because more computational effort can be placed on the image deconvolution problem and therefore accelerate convergence. Since the computational cost of solving the image deconvolution problem is relatively small, multiple image-based EM iterations do not significantly increase the overall reconstruction time. The proposed algorithm was evaluated using 2D simulations, as well as measured 3D data acquired on the high-resolution research tomograph. Results showed that bias reduction can be accelerated by interleaving multiple iterations of the image-based EM algorithm solving the resolution model problem, with a single EM iteration solving the tomographic problem. Significant improvements were observed particularly for voxels that were located on the boundaries between regions of high contrast within the object being imaged and for small regions of interest, where resolution recovery is usually more challenging. Minor differences were observed using the proposed nested algorithm, compared to the single iteration normally performed, when an optimal number of iterations are performed for each algorithm. However, using the proposed nested approach convergence is significantly accelerated enabling reconstruction using far fewer tomographic iterations (up to 70% fewer iterations for small regions). Nevertheless, the optimal number of nested image-based EM iterations is hard to be defined and it should be selected according to the given application.

The performance of monotonic and new non-monotonic gradient ascent reconstruction algorithms for high-resolution neuroreceptor PET imaging.

Iterative expectation maximization (EM) techniques have been extensively used to solve maximum likelihood (ML) problems in positron emission tomography (PET) image reconstruction. Although EM methods offer a robust approach to solving ML problems, they usually suffer from slow convergence rates. The ordered subsets EM (OSEM) algorithm provides significant improvements in the convergence rate, but it can cycle between estimates converging towards the ML solution of each subset. In contrast, gradient-based methods, such as the recently proposed non-monotonic maximum likelihood (NMML) and the more established preconditioned conjugate gradient (PCG), offer a globally convergent, yet equally fast, alternative to OSEM. Reported results showed that NMML provides faster convergence compared to OSEM; however, it has never been compared to other fast gradient-based methods, like PCG. Therefore, in this work we evaluate the performance of two gradient-based methods (NMML and PCG) and investigate their potential as an alternative to the fast and widely used OSEM. All algorithms were evaluated using 2D simulations, as well as a single [(11)C]DASB clinical brain dataset. Results on simulated 2D data show that both PCG and NMML achieve orders of magnitude faster convergence to the ML solution compared to MLEM and exhibit comparable performance to OSEM. Equally fast performance is observed between OSEM and PCG for clinical 3D data, but NMML seems to perform poorly. However, with the addition of a preconditioner term to the gradient direction, the convergence behaviour of NMML can be substantially improved. Although PCG is a fast convergent algorithm, the use of a (bent) line search increases the complexity of the implementation, as well as the computational time involved per iteration. Contrary to previous reports, NMML offers no clear advantage over OSEM or PCG, for noisy PET data. Therefore, we conclude that there is little evidence to replace OSEM as the algorithm of choice for many applications, especially given that in practice convergence is often not desired for algorithms seeking ML estimates.

A custom-built PET phantom design for quantitative imaging of printed distributions.

This note presents a practical approach to a custom-made design of PET phantoms enabling the use of digital radioactive distributions with high quantitative accuracy and spatial resolution. The phantom design allows planar sources of any radioactivity distribution to be imaged in transaxial and axial (sagittal or coronal) planes. Although the design presented here is specially adapted to the high-resolution research tomograph (HRRT), the presented methods can be adapted to almost any PET scanner. Although the presented phantom design has many advantages, a number of practical issues had to be overcome such as positioning of the printed source, calibration, uniformity and reproducibility of printing. A well counter (WC) was used in the calibration procedure to find the nonlinear relationship between digital voxel intensities and the actual measured radioactive concentrations. Repeated printing together with WC measurements and computed radiography (CR) using phosphor imaging plates (IP) were used to evaluate the reproducibility and uniformity of such printing. Results show satisfactory printing uniformity and reproducibility; however, calibration is dependent on the printing mode and the physical state of the cartridge. As a demonstration of the utility of using printed phantoms, the image resolution and quantitative accuracy of reconstructed HRRT images are assessed. There is very good quantitative agreement in the calibration procedure between HRRT, CR and WC measurements. However, the high resolution of CR and its quantitative accuracy supported by WC measurements made it possible to show the degraded resolution of HRRT brain images caused by the partial-volume effect and the limits of iterative image reconstruction.

Single scan parameterization of space-variant point spread functions in image space via a printed array: the impact for two PET/CT scanners.

Incorporation of a resolution model during statistical image reconstruction often produces images of improved resolution and signal-to-noise ratio. A novel and practical methodology to rapidly and accurately determine the overall emission and detection blurring component of the system matrix using a printed point source array within a custom-made Perspex phantom is presented. The array was scanned at different positions and orientations within the field of view (FOV) to examine the feasibility of extrapolating the measured point source blurring to other locations in the FOV and the robustness of measurements from a single point source array scan. We measured the spatially-variant image-based blurring on two PET/CT scanners, the B-Hi-Rez and the TruePoint TrueV. These measured spatially-variant kernels and the spatially-invariant kernel at the FOV centre were then incorporated within an ordinary Poisson ordered subset expectation maximization (OP-OSEM) algorithm and compared to the manufacturer's implementation using projection space resolution modelling (RM). Comparisons were based on a point source array, the NEMA IEC image quality phantom, the Cologne resolution phantom and two clinical studies (carbon-11 labelled anti-sense oligonucleotide [(11)C]-ASO and fluorine-18 labelled fluoro-l-thymidine [(18)F]-FLT). Robust and accurate measurements of spatially-variant image blurring were successfully obtained from a single scan. Spatially-variant resolution modelling resulted in notable resolution improvements away from the centre of the FOV. Comparison between spatially-variant image-space methods and the projection-space approach (the first such report, using a range of studies) demonstrated very similar performance with our image-based implementation producing slightly better contrast recovery (CR) for the same level of image roughness (IR). These results demonstrate that image-based resolution modelling within reconstruction is a valid alternative to projection-based modelling, and that, when using the proposed practical methodology, the necessary resolution measurements can be obtained from a single scan. This approach avoids the relatively time-consuming and involved procedures previously proposed in the literature.

Nonuniqueness in diffusion-based optical tomography.

A condition on nonuniqueness in optical tomography is stated. The main result applies to steady-state (dc) diffusion-based optical tomography, wherein we demonstrate that simultaneous unique recovery of diffusion and absorption coefficients cannot be achieved. A specific example of two images that give identical dc data is presented. If the refractive index is considered an unknown, then nonuniqueness also occurs in frequency-domain and time-domain optical tomography, if the underlying model of the diffusion approximation is employed.