Multi-material decomposition of spectral CT images via Fully Convolutional DenseNets.

BACKGROUND: Spectral computed tomography (CT) has the capability to resolve the energy levels of incident photons, which has the potential to distinguish different material compositions. Although material decomposition methods based on x-ray attenuation characteristics have good performance in dual-energy CT imaging, there are some limitations in terms of image contrast and noise levels. OBJECTIVE: This study focused on multi-material decomposition of spectral CT images based on a deep learning approach. METHODS: To classify and quantify different materials, we proposed a multi-material decomposition method via the improved Fully Convolutional DenseNets (FC-DenseNets). A mouse specimen was first scanned by spectral CT system based on a photon-counting detector with different energy ranges. We then constructed a training set from the reconstructed CT images for deep learning to decompose different materials. RESULTS: Experimental results demonstrated that the proposed multi-material decomposition method could more effectively identify bone, lung and soft tissue than the basis material decomposition based on post-reconstruction space in high noise levels. CONCLUSIONS: The new proposed approach yielded good performance on spectral CT material decomposition, which could establish guidelines for multi-material decomposition approaches based on the deep learning algorithm.

The Experimental Study on Geometric Calibration and Material Discrimination for In Vivo Dual-Energy CT Imaging.

Photon-counting detector (PCD) can identify absorption features in the multiple ranges of photon energies, which has a great potential in material discrimination. In this paper, we focused on in vivo dual-energy CT imaging to characterize different biomedical compositions. The precision of material decomposition in post-reconstruction space depends on the quality of reconstructed CT images; we used the locally linear embedding (LLE) based online geometric calibration method and GPU-based reconstruction toolbox to reconstruct high-quality CT images. Then, we performed the real experiment and studied materials decomposition with basis material model to discriminate soft tissue and cortical bone of small animal. Finally, the experimental results demonstrated that the proposed method could reconstruct small animal CT images with more slim structures and details, and improve the precision of materials decomposition in dual-energy CT imaging.

Convolutional Sparse Coding for Compressed Sensing CT Reconstruction.

Over the past few years, dictionary learning (DL)-based methods have been successfully used in various image reconstruction problems. However, the traditional DL-based computed tomography (CT) reconstruction methods are patch-based and ignore the consistency of pixels in overlapped patches. In addition, the features learned by these methods always contain shifted versions of the same features. In recent years, convolutional sparse coding (CSC) has been developed to address these problems. In this paper, inspired by several successful applications of CSC in the field of signal processing, we explore the potential of CSC in sparse-view CT reconstruction. By directly working on the whole image, without the necessity of dividing the image into overlapped patches in DL-based methods, the proposed methods can maintain more details and avoid artifacts caused by patch aggregation. With predetermined filters, an alternating scheme is developed to optimize the objective function. Extensive experiments with simulated and real CT data were performed to validate the effectiveness of the proposed methods. The qualitative and quantitative results demonstrate that the proposed methods achieve better performance than the several existing state-of-the-art methods.

The small animal material discrimination study based on equivalent monochromatic energy projection decomposition method with dual-energy CT system.

Material discrimination is an important application of dual-energy computed tomography (CT) techniques. Projection decomposition is a key problem for pre-reconstruction material discrimination. In this study, we focused on the pre-reconstruction space based on the photoelectric and Compton effect decomposition model to characterize different material components, and proposed an efficient method to calculate the projection decomposition coefficient. We converted the complex projection integral into a linear equation by calculating the equivalent monochromatic energy from the high and low energy spectrum. Meanwhile, we constructed a dual-energy CT system based on a photon-counting detector to take small animal scan and material discrimination analysis. Finally, the results of simulation and experimental study demonstrated the feasibility of our proposed new method, and explained the characteristics of photoelectric absorption and Compton scattering reconstruction images.

Monte Carlo Simulation for Polychromatic X-Ray Fluorescence Computed Tomography with Sheet-Beam Geometry.

X-ray fluorescence computed tomography (XFCT) based on sheet beam can save a huge amount of time to obtain a whole set of projections using synchrotron. However, it is clearly unpractical for most biomedical research laboratories. In this paper, polychromatic X-ray fluorescence computed tomography with sheet-beam geometry is tested by Monte Carlo simulation. First, two phantoms (A and B) filled with PMMA are used to simulate imaging process through GEANT 4. Phantom A contains several GNP-loaded regions with the same size (10 mm) in height and diameter but different Au weight concentration ranging from 0.3% to 1.8%. Phantom B contains twelve GNP-loaded regions with the same Au weight concentration (1.6%) but different diameter ranging from 1 mm to 9 mm. Second, discretized presentation of imaging model is established to reconstruct more accurate XFCT images. Third, XFCT images of phantoms A and B are reconstructed by filter back-projection (FBP) and maximum likelihood expectation maximization (MLEM) with and without correction, respectively. Contrast-to-noise ratio (CNR) is calculated to evaluate all the reconstructed images. Our results show that it is feasible for sheet-beam XFCT system based on polychromatic X-ray source and the discretized imaging model can be used to reconstruct more accurate images.

An Improved Total Variation Minimization Method Using Prior Images and Split-Bregman Method in CT Reconstruction.

Compressive Sensing (CS) theory has great potential for reconstructing Computed Tomography (CT) images from sparse-views projection data and Total Variation- (TV-) based CT reconstruction method is very popular. However, it does not directly incorporate prior images into the reconstruction. To improve the quality of reconstructed images, this paper proposed an improved TV minimization method using prior images and Split-Bregman method in CT reconstruction, which uses prior images to obtain valuable previous information and promote the subsequent imaging process. The images obtained asynchronously were registered via Locally Linear Embedding (LLE). To validate the method, two studies were performed. Numerical simulation using an abdomen phantom has been used to demonstrate that the proposed method enables accurate reconstruction of image objects under sparse projection data. A real dataset was used to further validate the method.

Fully 3D geometrical calibration for an X-ray grating-based imaging system.

The Precision of geometric parameters is the key requirement for grating-based phase-contrast CT and attenuation-based CT in cone-beam geometry. By extending our fan-beam geometric calibration work via Locally Linear Embedding (LLE) into the cone-beam case, here we calibrate the geometric parameters of our in-house phase-contrast CT system with attenuation projection data. Numerical and experimental studies show that the LLE-based calibration method is feasible to calibrate 8 parameters of the cone-beam geometry that improves reconstruction quality significantly.

A mixed reality approach for stereo-tomographic quantification of lung nodules.

To reduce the radiation dose and the equipment cost associated with lung CT screening, in this paper we propose a mixed reality based nodule measurement method with an active shutter stereo imaging system. Without involving hundreds of projection views and subsequent image reconstruction, we generated two projections of an iteratively placed ellipsoidal volume in the field of view and merging these synthetic projections with two original CT projections. We then demonstrated the feasibility of measuring the position and size of a nodule by observing whether projections of an ellipsoidal volume and the nodule are overlapped from a human observer's visual perception through the active shutter 3D vision glasses. The average errors of measured nodule parameters are less than 1 mm in the simulated experiment with 8 viewers. Hence, it could measure real nodules accurately in the experiments with physically measured projections.

X-ray CT geometrical calibration via locally linear embedding.

For X-ray computed tomography (CT), geometric calibration and rigid patient motion compensation are inter-related issues for optimization of image reconstruction quality. Non-calibrated system geometry and patient movement during a CT scan will result in streak-like, blurring and other artifacts in reconstructed images. In this paper, we propose a locally linear embedding based calibration approach to address this challenge under a rigid 2D object assumption and a more general way than what has been reported before. In this method, projections are linearly represented by up-sampled neighbors via locally linear embedding, and CT system parameters are iteratively estimated from projection data themselves. Numerical and experimental studies show that images reconstructed with calibrated parameters are in excellent agreement with the counterparts reconstructed with the true parameters.

A CT reconstruction approach from sparse projection with adaptive-weighted diagonal total-variation in biomedical application.

For lack of directivity in Total Variation (TV) which only uses x-coordinate and y-coordinate gradient transform as its sparse representation approach during the iteration process, this paper brought in Adaptive-weighted Diagonal Total Variation (AwDTV) that uses the diagonal direction gradient to constraint reconstructed image and adds associated weights which are expressed as an exponential function and can be adaptively adjusted by the local image-intensity diagonal gradient for the purpose of preserving the edge details, then using the steepest descent method to solve the optimization problem. Finally, we did two sets of numerical simulation and the results show that the proposed algorithm can reconstruct high-quality CT images from few-views projection, which has lower Root Mean Square Error (RMSE) and higher Universal Quality Index (UQI) than Algebraic Reconstruction Technique (ART) and TV-based reconstruction method.

A CT reconstruction algorithm based on non-aliasing Contourlet transform and compressive sensing.

Compressive sensing (CS) theory has great potential for reconstructing CT images from sparse-views projection data. Currently, total variation (TV-) based CT reconstruction method is a hot research point in medical CT field, which uses the gradient operator as the sparse representation approach during the iteration process. However, the images reconstructed by this method often suffer the smoothing problem; to improve the quality of reconstructed images, this paper proposed a hybrid reconstruction method combining TV and non-aliasing Contourlet transform (NACT) and using the Split-Bregman method to solve the optimization problem. Finally, the simulation results show that the proposed algorithm can reconstruct high-quality CT images from few-views projection using less iteration numbers, which is more effective in suppressing noise and artefacts than algebraic reconstruction technique (ART) and TV-based reconstruction method.

A CT Reconstruction Algorithm Based on L1/2 Regularization.

Computed tomography (CT) reconstruction with low radiation dose is a significant research point in current medical CT field. Compressed sensing has shown great potential reconstruct high-quality CT images from few-view or sparse-view data. In this paper, we use the sparser L1/2 regularization operator to replace the traditional L1 regularization and combine the Split Bregman method to reconstruct CT images, which has good unbiasedness and can accelerate iterative convergence. In the reconstruction experiments with simulation and real projection data, we analyze the quality of reconstructed images using different reconstruction methods in different projection angles and iteration numbers. Compared with algebraic reconstruction technique (ART) and total variance (TV) based approaches, the proposed reconstruction algorithm can not only get better images with higher quality from few-view data but also need less iteration numbers.

Material discrimination based on K-edge characteristics.

Spectral/multienergy CT employing the state-of-the-art energy-discriminative photon-counting detector can identify absorption features in the multiple ranges of photon energies and has the potential to distinguish different materials based on K-edge characteristics. K-edge characteristics involve the sudden attenuation increase in the attenuation profile of a relatively high atomic number material. Hence, spectral CT can utilize material K-edge characteristics (sudden attenuation increase) to capture images in available energy bins (levels/windows) to distinguish different material components. In this paper, we propose an imaging model based on K-edge characteristics for maximum material discrimination with spectral CT. The wider the energy bin width is, the lower the noise level is, but the poorer the reconstructed image contrast is. Here, we introduce the contrast-to-noise ratio (CNR) criterion to optimize the energy bin width after the K-edge jump for the maximum CNR. In the simulation, we analyze the reconstructed image quality in different energy bins and demonstrate that our proposed optimization approach can maximize CNR between target region and background region in reconstructed image.