High-dimensional data can be represented as a concise combination of explanatory data ingredients, which is referred to as the nature of sparsity. Compressed sensing is a general paradigm of sparsity-aware data acquisition to improve the functionality and lower the cost of measurement. For the compressed sensing fundamentally formulated as an underdetermined system of linear equations having a sparse solution, there have been provided theoretical underpinnings of random measurement and sparse reconstruction, as well as efficient sparse solvers based on convex relaxation. In imaging applications, reconstructed images are supposed to have sparse features, e.g., edges of objects. One can consistently derive practical algorithms for such image reconstruction by posing it as a linearly constrained convex optimization problem.
Recently, three new designs of CT scanners called low-dose CT, sparse-view CT, and interior CT (local CT) have been proposed, and, image reconstruction algorithms corresponding to each scanner design have been investigated actively. Among them, in this paper, we review image reconstruction in the interior CT, which measures truncated projection data by radiating x-rays only to a small region of interest (ROI) such as heart or breast to reduce patient dose or to simplify measurement. In particular, we explain two exact reconstruction approaches discovered in the late 2000's, which have been considered a breakthrough in this field. The first approach does not rely on the compressed sensing (CS), and the second approach is strongly based on the CS technique.
The purpose of this article is to review the recent development of compressed sensing magnetic resonance imaging (CS-MRI) and related techniques. After the brief review of basics of MR imaging, characteristics of new technologies such as CS-MRI, MR fingerprinting are discussed.
In most MR scanning, image reconstruction is executed by taking the inverse Fourier transform of acquired signal. However, image reconstruction in compressed sensing requires the choice of sparsifying transform function and algorithms for L1 norm minimization of cost function, therefore, the obtained image quality depends on those functions and reconstruction algorithms. In this paper, comparison of iterative shrinkage algorithm and ADMM (alternating direction method of multipliers) in compressed sensing of phase-varied MR images was performed. Simulation studies showed that the highest PSNR was obtained in ADMM combined with multi-scale sparsifying transform function.
Super resolution (SR) is a technique to estimate a high-resolution (HR) image from one or several low-resolution (LR) images. SR can be broadly classified into two families of methods: (i) The classical multi-frame super-resolution, and (ii) Example-based or learning-based super-resolution, which is also known as single-frame super-resolution. In the classical multi-frame SR, the HR image is reconstructed by combining subpixel-aligned multi-images (LR images). Since the reconstruction of HR image from LR images is often an ill-posed problem, we need to include some a prior knowledge or additional assumptions for reconstruction. If sparsity is used as a prior knowledge, the SR can be considered as one of compressed sensing techniques. In the learning-based single-frame SR, the HR image is estimated by learning correspondence between low- and high-resolution images from a database. In this paper, we introduce a multi-frame SR and a single-frame SR for medical image enhancement. Both of them use the sparsity as a prior knowledge.
Palpation allows physicians to assess change in the mechanical properties of tissue associated with the presence and development of disease. Magnetic resonance elastography (MRE) is an imaging technique for the noninvasive quantification of the mechanical properties of tissue using magnetic resonance imaging (MRI). Typical MRE system uses an external driver to create shear wave in the tissue, a phase-contrast MRI pulse sequence to take snapshots of temporally sequential shear wave pattern in the tissue and an inversion algorithm to calculate viscoelastic map from the shear wave snapshots. These three components are improved over the last 20 years and strongly dependent on each other. The improvement of MRE system have allowed for imaging various organs and have shown effectiveness for staging and differential diagnosis. This review describes the principles, technique, and clinical applications of MRE.
PET (positron emission tomography) is an important molecular imaging method using 511 keV gamma-rays emitted from annihilation of positron tracers. Recently many new developments in scintillator and fast photo-sensor have been investigated. Here the data acquisition and signal processing in PET are reviewed and we introduce multi-channel gamma-ray detector required for PET system.