Compressed Sensing (CS) is important in the field of image processing and signal processing, and CS-Magnetic Resonance Imaging (MRI) is used to reconstruct image from undersampled k-space data. Total Variation (TV) regularisation is a common technique to improve the sparsity of image, and the Alternating Direction Multiplier Method (ADMM) plays a key role in the variational image processing problem. This paper aims to improve the quality of MRI and shorten the reconstruction time. We consider MRI to solve a linear inverse problem, we convert it into a constrained optimization problem based on TV regularisation, then an accelerated ADMM is established. Through a series of theoretical derivations, we verify that the algorithm satisfies the convergence rate of O1/k under the condition that one objective function is quadratically convex and the other is strongly convex. We select five undersampled templates for testing in MRI experiment and compare it with other algorithms, experimental results show that our proposed method not only improves the running speed but also gives better reconstruction results.

Intravascular enhancement sign (IVES) at three-dimensional T1-weighted turbo spin echo (3D T1W TSE) sequence may be a simple hemodynamic maker. This study aims to investigate the association between IVES and features of intracranial atherosclerotic stenosis (ICAS).

Transaxial CT imaging is the main clinical imaging modality for the assessment of COVID-induced lung damage. However, this type of data does not quantify the functional properties of the lung. The objective is to provide non-invasive personalized cartographies of lung stiffness for long-COVID patients using MR elastography (MRE) and follow-up the evolution of this quantitative mapping over time.

Temporal encephaloceles (TEs) are herniations of cerebral parenchyma through structural defects in the floor of the middle cranial fossa. They are a relatively common, but only relatively recently identified potential cause of drug-resistant epilepsy. Uncontrolled epilepsy is associated with many negative long term health consequences including a heightened risk of death. The most effective treatment for drug-resistant epilepsy is surgery. One of the most predictive factors associated with successful surgery is identification of an abnormality on imaging. However, TEs can be difficult to detect and are often overlooked on neuroimaging studies. Improving our ability to accurately detect TEs by MRI is an important step in improving surgical outcomes in patients with drug-resistant epilepsy. We performed a review on existing imaging modalities for detecting TEs and report on our attempt to use a voxel-based morphometry (VBM) algorithm to detect TEs in T1-weighted MRIs of 81 patients from a database comprised of 25 patients with confirmed encephaloceles and 56 controls. Our program's sensitivity and specificity were compared to those of two neuroradiologists and two epileptologists using visualization during surgery as the gold standard. On average, the neuroradiologists and epileptologists had sensitivities of 41 % and 58 % and specificities of 81 % and 60 % while our VBM-based approach had sensitivities and specificities ranging from 11 % to 50 % and 0.2 % to 17 %, respectively. This work provides an overview of the different imaging modalities utilized in the detection of TEs and highlights the difficulties associated with their detection for both experienced physicians and cutting-edge computational methods. Our findings suggest that VBM-based methods could potentially be used to enhance clinicians' ability to detect TEs thereby facilitating surgical planning, improving surgical outcomes by allowing for more specific targeting, and bettering the long-term health and well-being of patients with drug-resistant epilepsy secondary to TEs.

There has been tremendous progress in MRI over the past 40+ years, driven by advances in technology as well as human ingenuity, with considerable impact in medicine. However, our understanding of how to account for, and interpret, MRI properties quantitatively lags behind these technical advances. This lack of understanding will limit our ability to make full use of quantitative metrics in the future, and much more work is needed to bridge this knowledge gap.

PROPELLER MRI has been shown effective for rigid motion compensation, while the performance of existing PROPELLER reconstruction methods critically depend on selecting a proper reference blade. In this work, we proposed a robust implementation for PROPELLER reconstruction, which was incorporated with different reference selection methods, including single blade reference (SBR), combined blades reference (CBR), grouped blades reference (GBR) and Pipe et al.'s revised method, which requires no blade reference (NBR). Both simulation and in vivo studies were performed to evaluate the precision and robustness of motion estimation for reference selection methods. In vivo data sets from 10 volunteers with instructed motion and 11 patients with random motion were collected and images were scored independently and blindly by two experienced radiologists. Both simulation and in vivo studies demonstrate that the four reference selection methods have similar performances according to visual inspection. In our tests, one iteration for the motion estimation can be sufficient for SBR, CBR, or GBR, and comparable to NBR in terms of image quality for clinical diagnosis. With two iterations, SBR, CBR, and GBR are comparable to NBR in terms of motion estimation precision. With our proposed PROPELLER reconstruction, reference selection is not critical for robust motion correction. NBR with no iterations and SBR, CBR, and GBR with two iterations are recommended for accurate motion correction.

In this letter to the editor, authors highlight the key role of data labeling in training AI models for medical imaging, discussing the complexities, resource demands, costs, and the relevance of quality control in the labeling process including the potential and limitations of AI tools for automated labeling. The article underlines that labeling quality is essential for the accuracy of AI models and the safety of their clinical applications, highlighting the legal responsibilities of labelers in cases where improper labeling leads to AI errors.

Hyperpolarized Xenon MRI (HPG MRI) has been studied for its potential use in assessing lung function in patients with cystic fibrosis (CF) and in patients with asthma. We present a case of a man with overlapping cystic fibrosis and allergic asthma with severe obstructive lung disease in which spirometry and computed topography (CT) imaging was unable to determine the primary cause for his uncontrolled symptoms. HPG MRI was used to guide a tissue biopsy and determine the primary driver to be allergic asthma. After starting targeted therapy for severe asthma, his symptoms have greatly improved.

BUDA-cEPI has been shown to achieve high-quality, high-resolution diffusion magnetic resonance imaging (dMRI) with fast acquisition time, particularly when used in conjunction with S-LORAKS reconstruction. However, this comes at a cost of more complex reconstruction that is computationally prohibitive. In this work we develop rapid reconstruction pipeline for BUDA-cEPI to pave the way for its deployment in routine clinical and neuroscientific applications. The proposed reconstruction includes the development of ML-based unrolled reconstruction as well as rapid ML-based B0 and eddy current estimations that are needed. The architecture of the unroll network was designed so that it can mimic S-LORAKS regularization well, with the addition of virtual coil channels.

This study introduces GraFMRI, a novel framework designed to address the challenges of reconstructing high-quality MRI images from undersampled k-space data. Traditional methods often suffer from noise amplification and loss of structural detail, leading to suboptimal image quality. GraFMRI leverages Graph Neural Networks (GNNs) to transform multi-modal MRI data (T1, T2, PD) into a graph-based representation, enabling the model to capture intricate spatial relationships and inter-modality dependencies.

Medical Imaging (MI) tasks, such as accelerated parallel Magnetic Resonance Imaging (MRI), often involve reconstructing an image from noisy or incomplete measurements. This amounts to solving ill-posed inverse problems, where a satisfactory closed-form analytical solution is not available. Traditional methods such as Compressed Sensing (CS) in MRI reconstruction can be time-consuming or prone to obtaining low-fidelity images. Recently, a plethora of Deep Learning (DL) approaches have demonstrated superior performance in inverse-problem solving, surpassing conventional methods. In this study, we propose vSHARP (variable Splitting Half-quadratic ADMM algorithm for Reconstruction of inverse Problems), a novel DL-based method for solving ill-posed inverse problems arising in MI. vSHARP utilizes the Half-Quadratic Variable Splitting method and employs the Alternating Direction Method of Multipliers (ADMM) to unroll the optimization process. For data consistency, vSHARP unrolls a differentiable gradient descent process in the image domain, while a DL-based denoiser, such as a U-Net architecture, is applied to enhance image quality. vSHARP also employs a dilated-convolution DL-based model to predict the Lagrange multipliers for the ADMM initialization. We evaluate vSHARP on tasks of accelerated parallel MRI Reconstruction using two distinct datasets and on accelerated parallel dynamic MRI Reconstruction using another dataset. Our comparative analysis with state-of-the-art methods demonstrates the superior performance of vSHARP in these applications.

Atrial fibrillation (AF) is the most prevalent arrhythmia in world-wild places and is associated with the development of severe secondary complications such as heart failure and stroke. Emerging evidence shows that the modified hemodynamic environment associated with AF can cause altered flow patterns in left atrial and even systemic blood associated with left atrial appendage thrombosis. Recent advances in magnetic resonance imaging (MRI) allow for the comprehensive visualization and quantification of in vivo aortic flow pattern dynamics. In particular, the technique of 4- dimensional flow MRI (4D flow MRI) offers the opportunity to derive advanced hemodynamic measures such as velocity, vortex, endothelial cell activation potential, and kinetic energy. This review introduces 4D flow MRI for blood flow visualization and quantification of hemodynamic metrics in the setting of AF, with a focus on AF and associated secondary complications.

The identification and measurement of aortic aneurysms is an important clinical problem. While specialized high-resolution 3D CMR sequences allow detailed aortic assessment, they are time-consuming which limits their use in screening routine cardiac scans and in population studies.

Degeneration of white matter (WM) microstructure in the central nervous system is characteristic of many neurodegenerative conditions. Previous research indicates that axonal degeneration visible in ex vivo electron microscopy (EM) photomicrographs precede the onset of clinical symptoms. Measuring WM microstructural features, such as axon diameter and packing fraction, currently require these highly invasive methods of analysis and it is therefore of great importance to develop methods for in vivo measurements. Diffusion weighted Magnetic Resonance Imaging (MRI) is a non-invasive method which can be used in conjunction with temporal diffusion spectroscopy (TDS) and an oscillating gradient spin echo (OGSE) pulse sequence to probe micron-scale structures within neural tissue. The current experiment aims to compare axon diameter measurements, mean effective axon diameter (AxD¯), and packing fractions calculated from EM histopathological analysis and inferred values from MR images. Mathematical models of axon diameters used for analysis include the ActiveAx Frequency-Dependent Extra-Axonal Diffusion (AAD) model and the AxCaliber Frequency-Dependent Extra-Axonal Diffusion (ACD) model using ROI (Region of Interest) based analysis (RBA) and voxel-based analysis (VBA), respectively. Overall, it was observed that MRI inferred WM microstructural parameters overestimate those calculated from EM. This may be attributable to tissue shrinkage during EM dehydration, the sensitivity of MR pulse sequences to larger diameter axons, and/or inaccurate model assumptions. The results of the current study provide a means to characterize the precision and accuracy of RBA-ACD and VBA-AAD OGSE-TDS and highlight the need for further research investigating the relationship between ex vivo MRI and EM, with the goal of reaching in vivo MRI.

Magnetic resonance imaging (MRI) is widely used in clinical diagnosis as a safe, non-invasive, high-resolution medical imaging technology, but long scanning time has been a major challenge for this technology. The undersampling reconstruction method has become an important technical means to accelerate MRI by reducing the data sampling rate while maintaining high-quality imaging. However, traditional undersampling reconstruction techniques such as compressed sensing mainly rely on relatively single sparse or low-rank prior information to reconstruct the image, which has limitations in capturing the comprehensive features of images, resulting in the insufficient performance of the reconstructed image in terms of details and key information. In this paper, we propose a deep plug-and-play multiple complementary priors MRI reconstruction model, which combines traditional low-rank matrix recovery model methods and deep learning methods, and integrates global, local and nonlocal priors to improve reconstruction quality. Specifically, we capture the global features of the image through the matrix nuclear norm, and use the deep convolutional neural network denoiser Swin-Conv-UNet (SCUNet) and block-matching and 3-D filtering (BM3D) algorithm to preserve the local details and structural texture of the image, respectively. In addition, we utilize an efficient half-quadratic splitting (HQS) algorithm to solve the proposed model. The experimental results show that our proposed method has better reconstruction ability than the existing popular methods in terms of visual effects and numerical results.

In fMRI, capturing brain activity during a task is dependent on how quickly the k-space arrays for each volume image are obtained. Acquiring the full k-space arrays can take a considerable amount of time. Under-sampling k-space reduces the acquisition time, but results in aliased, or "folded," images after applying the inverse Fourier transform (IFT). GeneRalized Autocalibrating Partial Parallel Acquisition (GRAPPA) and SENSitivity Encoding (SENSE) are parallel imaging techniques that yield reconstructed images from subsampled arrays of k-space. With GRAPPA operating in the spatial frequency domain and SENSE in image space, these techniques have been separate but can be merged to reconstruct the subsampled k-space arrays more accurately. Here, we propose a Bayesian approach to this merged model where prior distributions for the unknown parameters are assessed from a priori k-space arrays. The prior information is utilized to estimate the missing spatial frequency values, unalias the voxel values from the posterior distribution, and reconstruct into full field-of-view images. Our Bayesian technique successfully reconstructed simulated and experimental fMRI time series with no aliasing artifacts while decreasing temporal variation and increasing task detection power.

Mesial temporal sclerosis (MTS) is the most common pathology associated with drug-resistant mesial temporal lobe epilepsy (mTLE) in adults. Most atrophic hippocampi can be identified using MRI based on standard epilepsy protocols; however, difficulties can arise in cases where sclerotic changes in the hippocampus are subtle or non-epilepsy-specific protocols have been implemented. In such cases, quantitative methods, such as T1-weighted axial series MRIs, are valuable additional tools to complement epilepsy-specific protocols. In the current study, we applied machine learning (ML) techniques to the analysis of brain regions of interest (ROIs), including the hippocampus, thalamus, and cortical areas, to enhance the accuracy of lesion lateralization in MRI.

Common-mode currents can degrade the RF coil performance and introduce potential safety hazards in MRI. Baluns are the standard method to suppress these undesired common-mode currents. Specifically, floating baluns are preferred in many applications because they are removable, allow post-installation adjustment and avoid direct soldering on the cable. However, floating baluns are typically bulky to achieve excellent common-mode suppression, taking up valuable space in the MRI bore. This is particularly severe for multi-nuclear MRI/MRS applications, as two RF systems exist. In this work, we present a novel dual-tuned floating balun that is fully removable, does not require any physical connection to the coaxial cable, and has a significantly reduced footprint. The floating design employs an inductive coupling between the cable solenoid and a floating solenoid resonator rather than a direct physical connection. Unlike the previous floating solenoid balun, this balun employs a two-layer design further to improve the mutual coupling between the two solenoids. A pole-insertion method is used to suppress common-mode currents at two user-selectable frequencies simultaneously. Bench testing of the fabricated device at 7 T demonstrated high common-mode rejection ratios at Larmor frequencies of both H and Na, even with a compact dimension (diameter 18 mm and length 12 mm). This balun's removable, compact, and multi-resonant nature enables light-weighting, allows more coil elements, and improves cable management for advanced multi-nuclear MRI/MRS systems.

In Magnetic Resonance Imaging (MRI), the sequential acquisition of raw complex-valued image data in Fourier space, also known as k-space, results in extended examination times. To speed up the MRI scans, k-space data are usually undersampled and processed using numerical techniques such as compressed sensing (CS). While the majority of CS-MRI algorithms primarily focus on magnitude images due to their significant diagnostic value, the phase components of complex-valued MRI images also hold substantial importance for clinical diagnosis, including neurodegenerative diseases. In this work, complex-valued MRI reconstruction is studied with a focus on the simultaneous reconstruction of both magnitude and phase images. The proposed algorithm is based on the nonsubsampled contourlet transform (NSCT) technique, which offers shift invariance in images. Instead of directly transforming the complex-valued image into the NSCT domain, we introduce a wavelet transform within the NSCT domain, reducing the size of the sparsity of coefficients. This two-level hierarchical constraint (HC) enforces sparse representation of complex-valued images for CS-MRI implementation. The proposed HC is seamlessly integrated into a proximal algorithm simultaneously. Additionally, to effectively minimize the artifacts caused by sub-sampling, thresholds related to different sub-bands in the HC are applied through an alternating optimization process. Experimental results show that the novel method outperforms existing CS-MRI techniques in phase-regularized complex-valued image reconstructions.

A newly developed magnetic resonance imaging (MRI) approach is based on "Radiowave amplification by the stimulated emission of radiation" (RASER). RASER MRI potentially allows for higher resolution, is inherently background-free, and does not require radio-frequency excitation. However, RASER MRI can be "nearly unusable" as heavy distortions from nonlinear effects can occur. In this work, we show that deep learning (DL) reduces such artifacts in RASER images. We trained a two-step DL pipeline on purely synthetic data, which was generated based on a previously published, theoretical model for RASER MRI. A convolutional neural network was trained on 630'000 1D RASER projections, and a U-net on 2D random images. The DL pipeline generalizes well when applied from synthetic to experimental RASER MRI data.