Essam A. Rashed

Multiresolution mammogram analysis in multilevel decomposition

A multiresolution analysis system for interpreting digital mammograms is proposed and tested. This system is based on using fractional amount of biggest wavelets coefficients in multilevel decomposition. A set of real labeled database is used in evaluating the proposed system. The evaluation results show that the system has a remarkably high efficiency compared by other systems known till present especially in the area of distinguishing between benign and malignant tumors.

Statistical image reconstruction from limited projection data with intensity priors

The radiation dose generated from x-ray computed tomography (CT) scans and its responsibility for increasing the risk of malignancy became a major concern in the medical imaging community. Accordingly, investigating possible approaches for image reconstruction from low-dose imaging protocols, which minimize the patient radiation exposure without affecting image quality, has become an issue of interest. Statistical reconstruction (SR) methods are known to achieve a superior image quality compared with conventional analytical methods. Effective physical noise modeling and possibilities of incorporating priors in the image reconstruction problem are the main advantages of the SR methods. Nevertheless, the high computation cost limits its wide use in clinical scanners. This paper presents a framework for SR in x-ray CT when the angular sampling rate of the projection data is low. The proposed framework is based on the fact that, in many CT imaging applications, some physical and anatomical structures and the corresponding attenuation information of the scanned object can be a priori known. Therefore, the x-ray attenuation distribution in some regions of the object can be expected prior to the reconstruction. Under this assumption, the proposed method is developed by incorporating this prior information into the image reconstruction objective function to suppress streak artifacts. We limit the prior information to only a set of intensity values that represent the average intensity of the normal and expected homogeneous regions within the scanned object. This prior information can be easily computed in several x-ray CT applications. Considering the theory of compressed sensing, the objective function is formulated using the ℓ(1) norm distance between the reconstructed image and the available intensity priors. Experimental comparative studies applied to simulated data and real data are used to evaluate the proposed method. The comparison indicates a significant improvement in image quality when the proposed method is used.

Image reconstruction for sparse-view CT and interior CT— introduction to compressed sensing and differentiated backprojection

New designs of future computed tomography (CT) scanners called sparse-view CT and interior CT have been considered in the CT community. Since these CTs measure only incomplete projection data, a key to put these CT scanners to practical use is a development of advanced image reconstruction methods. After 2000, there was a large progress in this research area briefly summarized as follows. In the sparse-view CT, various image reconstruction methods using the compressed sensing (CS) framework have been developed towards reconstructing clinically feasible images from a reduced number of projection data. In the interior CT, several novel theoretical results on solution uniqueness and solution stability have been obtained thanks to the discovery of a new class of reconstruction methods called differentiated backprojection (DBP). In this paper, we mainly review this progress including mathematical principles of the CS image reconstruction and the DBP image reconstruction for readers unfamiliar with this area. We also show some experimental results from our past research to demonstrate that this progress is not only theoretically elegant but also works in practical imaging situations.

Sparsity-constrained three-dimensional image reconstruction for C-arm angiography

X-ray C-arm is an important imaging tool in interventional radiology, road-mapping and radiation therapy because it provides accurate descriptions of vascular anatomy and therapeutic end point. In common interventional radiology, the C-arm scanner produces a set of two-dimensional (2D) X-ray projection data obtained with a detector by rotating the scanner gantry around the patient. Unlike conventional fluoroscopic imaging, three-dimensional (3D) C-arm computed tomography (CT) provides more accurate cross-sectional images, which are helpful for therapy planning, guidance and evaluation in interventional radiology. However, 3D vascular imaging using the conventional C-arm fluoroscopy encounters some geometry challenges. Inspired by the theory of compressed sensing, we developed an image reconstruction algorithm for conventional angiography C-arm scanners. The main challenge in this image reconstruction problem is the projection data limitations. We consider a small number of views acquired from a short rotation orbit with offset scan geometry. The proposed method, called sparsity-constrained angiography (SCAN), is developed using the alternating direction method of multipliers, and the results obtained from simulated and real data are encouraging. SCAN algorithm provides a framework to generate 3D vascular images using the conventional C-arm scanners in lower cost than conventional 3D imaging scanners.

Row-action image reconstruction algorithm using ℓ p -norm distance to a reference image

This work investigates the problem of image reconstruction from small number of projection views in x-ray computed tomography (CT) imaging. The number of acquired projection views has a large influence on accuracy and stability of the image reconstruction problem. However, measuring the projection data over small number of views leads to a patient dose reduction and/or minimization of imaging time which become a principal target in many clinical applications. The presented work aims to develop a row-action type reconstruction algorithm that include a priori known information extracted from a reference image. The proposed method is based on the fact that, in many CT imaging applications, some physical and anatomical structures and the corresponding attenuation information of the scanned object can be a priori known. The main idea is to include a distance function consisting of ℓp norm of the reconstructed image into the cost function for image reconstruction. The constrained minimization problem is then transferred to the corresponding non-constrained maximization dual problem using Lagrangian duality. Experimental results indicate that the proposed reconstruction algorithms can effectively reduce the streak artifacts when a simple reference image is used.

Probabilistic atlas prior for CT image reconstruction

BACKGROUND AND OBJECTIVES In computed tomography (CT), statistical iterative reconstruction (SIR) approaches can produce images of higher quality compared to the conventional analytical methods such as filtered backprojection (FBP) algorithm. Effective noise modeling and possibilities to incorporate priors in the image reconstruction problem are the main advantages that lead to continuous development of SIR methods. Oriented by low-dose CT requirements, several methods are recently developed to obtain a high-quality image reconstruction from down-sampled or noisy projection data. In this paper, a new prior information obtained from probabilistic atlas is proposed for low-dose CT image reconstruction. METHODS The proposed approach consists of two main phases. In learning phase, a dataset of images obtained from different patients is used to construct a 3D atlas with Laplacian mixture model. The expectation maximization (EM) algorithm is used to estimate the mixture parameters. In reconstruction phase, prior information obtained from the probabilistic atlas is used to construct the cost function for image reconstruction. RESULTS We investigate the low-dose imaging by considering the reduction of X-ray beam intensity and by acquiring the projection data through a small number of views or limited view angles. Experimental studies using simulated data and chest screening CT data demonstrate that the probabilistic atlas prior is a practically promising approach for the low-dose CT imaging. CONCLUSIONS The prior information obtained from probabilistic atlas constructed from earlier scans of different patients is useful in low-dose CT imaging.

Three-dimensional angiography using mobile C-arm with IMU sensor attached: Initial study

Three-dimensional (3D) computed tomography (CT) imaging is becoming an essential demand in several clinical procedures. Mobile C-arm is a useful imaging tool for image-guided interventional radiology. C-arm systems are provided with X-ray image intensifier (XRII) or flat-panel detectors. Essentially, C-arm CT systems requires scanners with flat-panel detectors for its ability to provide homogenous image quality and improve the resolution of low-contrast subjects compared to those equipped with XRII. However, C-arm systems with XRIIs are widely used in several interventional procedures. Such systems can provide a high quality two-dimensional (2D) fluoroscopic images that facilitates minimal invasive surgery. However, it is unable to provide depth information for 3D imaging due to several factors. First, the gantry of XRII-based C-arms is usually operated manually, where the rotation angle is determined using printed angle scale attached to the scanner gantry. Second, the gantry orbital rotation is normally limited to angular range less than theoretically required for exact 3D reconstruction. Third, considering the offset-scan geometry, which is common configuration in mobile C-arm with XRII, the number of rays passing through field-of-view (FOV) is limited. In this paper, we develop a 3D angiographic imaging system using commercial C-arm system equipped with XRII. First, an in-house made gantry rotation unit is developed to control the scanner orbital rotation. Second, the gantry rotation is traced using inertial measurement unit (IMU) sensor attached to the scanner gantry. Geometry information obtained from IMU sensor are used to define the gantry position in the 3D space and synchronized with detector measurements. The SCAN algorithm is used for the 3D reconstruction and achieved results are of high quality.

GPR background removal using a directional total variation minimisation approach

Ground penetrating radar (GPR) is a leading geophysical subsurface imaging tool for various purposes. This efficiency, however, is compromised by the interference of different types of noise. Background noise (clutter) is one of the nagging types of noise that undermines the high-resolution imaging capabilities of GPR. This study presents the experience of applying a directional total variation minimisation (DTVM) filter to attenuate clutter in GPR data. The application of DTVM to both synthetic and field GPR data proves its great capability to attenuate clutter without affecting the features of interest of a GPR section. The results also show that the proposed DTVM method affords superior image quality than both the most commonly used and the most recently published background removal techniques.

Towards high-resolution synchrotron radiation imaging with statistical iterative reconstruction.

Synchrotron radiation (SR) X-ray micro-computed tomography (CT) is an effective imaging modality for high-resolution investigation of small objects, with several applications in medicine, biology and industry. However, the limited size of the detector field of view (FOV) restricts the sample dimensions to only a few millimeters. When the sample size is larger than the FOV, images reconstructed using conventional methods suffer from DC-shift and low-frequency artifacts. This classical problem is known as the local tomography or the interior problem. In this paper, a statistical iterative reconstruction method is introduced to eliminate image artifacts resulting from the local tomography. The proposed method, which can be used in several SR imaging applications, enables high-resolution SR imaging with superior image quality compared with conventional methods. Real data obtained from different SR micro-CT applications are used to evaluate the proposed method. Results indicate a noteworthy quality improvement in the image reconstructed from the local tomography measurements.

Practical statistical models for region-of-interest tomographic reconstruction and long object problem

This paper deals with reconstructing a small region- of-interest (ROI) of an object from non-truncated or truncated projection data by using statistical (iterative) methods. The imaging situations which we consider here can be classified into the following two scenarios. The first scenario is the case where non-truncated projection data is measured but only a small ROI needs to be reconstructed. The second scenario is the case where only truncated projection data passing through a ROI is measured and only the ROI needs to be reconstructed. When we blindly apply statistical methods to such cases, as described in the literature, the image matrix during the iteration must be large enough to contain the whole object (not only the ROI) even if the ROI to be reconstructed is small. This significantly increases the computational cost (approximately by a factor of the area of the whole object divided by the area of the ROI). Solutions to this problem have been investigated only very recently. We develop practical statistical models for the ROI reconstruction problem under the assumption that an initial estimate of the ROI image by an analytical method such as FBP or DBP is available. Thanks to a more rigorous treatment compared to the previous work, the proposed models are more accurate leading to significantly better noise properties as we demonstrate in the simulation study. Also, extending the proposed models to the 3D long object problem is discussed.