Non‐uniformly under‐sampled multi‐dimensional spectroscopic imaging in vivo: maximum entropy versus compressed sensing reconstruction

The four‐dimensional (4D) echo‐planar correlated spectroscopic imaging (EP‐COSI) sequence allows for the simultaneous acquisition of two spatial (k_y, k_x) and two spectral (t₂, t₁) dimensions in vivo in a single recording. However, its scan time is directly proportional to the number of increments in the k_y and t₁ dimensions, and a single scan can take 20–40 min using typical parameters, which is too long to be used for a routine clinical protocol. The present work describes efforts to accelerate EP‐COSI data acquisition by application of non‐uniform under‐sampling (NUS) to the k_y–t₁ plane of simulated and in vivo EP‐COSI datasets then reconstructing missing samples using maximum entropy (MaxEnt) and compressed sensing (CS). Both reconstruction problems were solved using the Cambridge algorithm, which offers many workflow improvements over other l₁‐norm solvers. Reconstructions of retrospectively under‐sampled simulated data demonstrate that the MaxEnt and CS reconstructions successfully restore data fidelity at signal‐to‐noise ratios (SNRs) from 4 to 20 and 5× to 1.25× NUS. Retrospectively and prospectively 4× under‐sampled 4D EP‐COSI in vivo datasets show that both reconstruction methods successfully remove NUS artifacts; however, MaxEnt provides reconstructions equal to or better than CS. Our results show that NUS combined with iterative reconstruction can reduce 4D EP‐COSI scan times by 75% to a clinically viable 5 min in vivo, with MaxEnt being the preferred method. Copyright © 2013 John Wiley & Sons, Ltd.     

Accelerated three‐dimensional cine phase contrast imaging using randomly undersampled echo planar imaging with compressed sensing reconstruction

The aim of this study was to implement and evaluate an accelerated three‐dimensional (3D) cine phase contrast MRI sequence by combining a randomly sampled 3D k‐space acquisition sequence with an echo planar imaging (EPI) readout. An accelerated 3D cine phase contrast MRI sequence was implemented by combining EPI readout with randomly undersampled 3D k‐space data suitable for compressed sensing (CS) reconstruction. The undersampled data were then reconstructed using low‐dimensional structural self‐learning and thresholding (LOST). 3D phase contrast MRI was acquired in 11 healthy adults using an overall acceleration of 7 (EPI factor of 3 and CS rate of 3). For comparison, a single two‐dimensional (2D) cine phase contrast scan was also performed with sensitivity encoding (SENSE) rate 2 and approximately at the level of the pulmonary artery bifurcation. The stroke volume and mean velocity in both the ascending and descending aorta were measured and compared between two sequences using Bland–Altman plots. An average scan time of 3 min and 30 s, corresponding to an acceleration rate of 7, was achieved for 3D cine phase contrast scan with one direction flow encoding, voxel size of 2 × 2 × 3 mm³, foot–head coverage of 6 cm and temporal resolution of 30 ms. The mean velocity and stroke volume in both the ascending and descending aorta were statistically equivalent between the proposed 3D sequence and the standard 2D cine phase contrast sequence. The combination of EPI with a randomly undersampled 3D k‐space sampling sequence using LOST reconstruction allows a seven‐fold reduction in scan time of 3D cine phase contrast MRI without compromising blood flow quantification. Copyright © 2014 John Wiley & Sons, Ltd.     

q‐Space diffusion MRI (QSI) of the disease progression in the spinal cords of the Long Evans shaker: diffusion time and apparent anisotropy

q‐Space diffusion MRI (QSI) was used to study the spinal cords of Long Evans shaker (les) rats, a model of dysmyelination, and their age‐matched controls at different maturation stages. Diffusion was measured parallel and perpendicular to the fibers of the spinal cords of the two groups and at different diffusion times. The results showed that QSI is able to detect the dysmyelination process that occurs in this model in the different stages of the disease. The differences in the diffusion characteristics of the spinal cords of the two groups were found to be larger when the diffusion time was increased from 22 to 100 ms. We found that the radial mean displacement is a much better parameter than the QSI fractional anisotropy (FA) to document the differences between the two groups. We observed that the degree of myelination affects the diffusion characteristics of the tissues, but has a smaller effect on FA. All of the extracted diffusion parameters that are affected by the degree of myelination are affected in a diffusion time‐dependent fashion, suggesting that the terms apparent anisotropy, apparent fractional anisotropy and even apparent root‐mean‐square displacement (rmsD) are more appropriate. Copyright © 2013 John Wiley & Sons, Ltd.     

A model‐based reconstruction for undersampled radial spin‐echo DTI with variational penalties on the diffusion tensor

Radial spin‐echo diffusion imaging allows motion‐robust imaging of tissues with very low T₂ values like articular cartilage with high spatial resolution and signal‐to‐noise ratio (SNR). However, in vivo measurements are challenging, due to the significantly slower data acquisition speed of spin‐echo sequences and the less efficient k‐space coverage of radial sampling, which raises the demand for accelerated protocols by means of undersampling. This work introduces a new reconstruction approach for undersampled diffusion‐tensor imaging (DTI). A model‐based reconstruction implicitly exploits redundancies in the diffusion‐weighted images by reducing the number of unknowns in the optimization problem and compressed sensing is performed directly in the target quantitative domain by imposing a total variation (TV) constraint on the elements of the diffusion tensor. Experiments were performed for an anisotropic phantom and the knee and brain of healthy volunteers (three and two volunteers, respectively). Evaluation of the new approach was conducted by comparing the results with reconstructions performed with gridding, combined parallel imaging and compressed sensing and a recently proposed model‐based approach. The experiments demonstrated improvements in terms of reduction of noise and streaking artifacts in the quantitative parameter maps, as well as a reduction of angular dispersion of the primary eigenvector when using the proposed method, without introducing systematic errors into the maps. This may enable an essential reduction of the acquisition time in radial spin‐echo diffusion‐tensor imaging without degrading parameter quantification and/or SNR. Copyright © 2015 John Wiley & Sons, Ltd.     

A data‐driven approach to optimising the encoding for multi‐shell diffusion MRI with application to neonatal imaging

Diffusion MRI has the potential to provide important information about the connectivity and microstructure of the human brain during normal and abnormal development, noninvasively and in vivo. Recent developments in MRI hardware and reconstruction methods now permit the acquisition of large amounts of data within relatively short scan times. This makes it possible to acquire more informative multi‐shell data, with diffusion sensitisation applied along many directions over multiple b‐value shells. Such schemes are characterised by the number of shells acquired, and the specific b‐value and number of directions sampled for each shell. However, there is currently no clear consensus as to how to optimise these parameters. In this work, we propose a means of optimising multi‐shell acquisition schemes by estimating the information content of the diffusion MRI signal, and optimising the acquisition parameters for sensitivity to the observed effects, in a manner agnostic to any particular diffusion analysis method that might subsequently be applied to the data. This method was used to design the acquisition scheme for the neonatal diffusion MRI sequence used in the developing Human Connectome Project (dHCP), which aims to acquire high quality data and make it freely available to the research community. The final protocol selected by the algorithm, and currently in use within the dHCP, consists of 20 b=0 images and diffusion‐weighted images at b = 400, 1000 and 2600 s/mm² with 64, 88 and 128 directions per shell, respectively.     

A radial sampling strategy for uniform k‐space coverage with retrospective respiratory gating in 3D ultrashort‐echo‐time lung imaging

The purpose of this work was to develop a 3D radial‐sampling strategy which maintains uniform k‐space sample density after retrospective respiratory gating, and demonstrate its feasibility in free‐breathing ultrashort‐echo‐time lung MRI. A multi‐shot, interleaved 3D radial sampling function was designed by segmenting a single‐shot trajectory of projection views such that each interleaf samples k‐space in an incoherent fashion. An optimal segmentation factor for the interleaved acquisition was derived based on an approximate model of respiratory patterns such that radial interleaves are evenly accepted during the retrospective gating. The optimality of the proposed sampling scheme was tested by numerical simulations and phantom experiments using human respiratory waveforms. Retrospectively, respiratory‐gated, free‐breathing lung MRI with the proposed sampling strategy was performed in healthy subjects. The simulation yielded the most uniform k‐space sample density with the optimal segmentation factor, as evidenced by the smallest standard deviation of the number of neighboring samples as well as minimal side‐lobe energy in the point spread function. The optimality of the proposed scheme was also confirmed by minimal image artifacts in phantom images. Human lung images showed that the proposed sampling scheme significantly reduced streak and ring artifacts compared with the conventional retrospective respiratory gating while suppressing motion‐related blurring compared with full sampling without respiratory gating. In conclusion, the proposed 3D radial‐sampling scheme can effectively suppress the image artifacts due to non‐uniform k‐space sample density in retrospectively respiratory‐gated lung MRI by uniformly distributing gated radial views across the k‐space. Copyright © 2016 John Wiley & Sons, Ltd.     

Comparison of different compressed sensing algorithms for low SNR 19F MRI applications—Imaging of transplanted pancreatic islets and cells labeled with perfluorocarbons

Transplantation of pancreatic islets is a possible treatment option for patients suffering from Type I diabetes. In vivo imaging of transplanted islets is important for assessment of the transplantation site and islet distribution. Thanks to its high specificity, the absence of intrinsic background signal in tissue and its potential for quantification, ¹⁹F MRI is a promising technique for monitoring the fate of transplanted islets in vivo. In order to overcome the inherent low sensitivity of ¹⁹F MRI, leading to long acquisition times with low signal‐to‐noise ratio (SNR), compressed sensing (CS) techniques are a valuable option. We have validated and compared different CS algorithms for acceleration of ¹⁹F MRI acquisition in a low SNR regime using pancreatic islets labeled with perfluorocarbons both in vitro and in vivo. Using offline simulation on both in vitro and in vivo low SNR fully sampled ¹⁹F MRI datasets of labeled islets, we have shown that CS is effective in reducing the image acquisition time by a factor of three to four without seriously affecting SNR, regardless of the particular algorithms used in this study, with the exception of CoSaMP. Using CS, signals can be detected that might have been missed by conventional ¹⁹F MRI. Among different algorithms (SPARSEMRI, OMMP, IRWL1, Two‐level and CoSAMP), the two‐level l₁ method has shown the best performance if computational time is taken into account. We have demonstrated in this study that different existing CS algorithms can be used effectively for low SNR ¹⁹F MRI. An up to fourfold gain in SNR/scan time could be used either to reduce the scan time, which is beneficial for clinical and translational applications, or to increase the number of averages, to potentially detect otherwise undetected signal when compared with conventional ¹⁹F MRI acquisitions. Potential applications in the field of cell therapy have been demonstrated.     

The impact of edema and fiber crossing on diffusion MRI metrics assessed in an ex vivo nerve phantom: Multi‐tensor model vs. diffusion orientation distribution function

Diffusion tensor imaging (DTI) has been employed for over 2 decades to noninvasively quantify central nervous system diseases/injuries. However, DTI is an inadequate simplification of diffusion modeling in the presence of coexisting inflammation, edema and crossing nerve fibers. We employed a tissue phantom using fixed mouse trigeminal nerves coated with various amounts of agarose gel to mimic crossing fibers in the presence of vasogenic edema. Diffusivity measures derived by DTI and diffusion basis spectrum imaging (DBSI) were compared at increasing levels of simulated edema and degrees of fiber crossing. Furthermore, we assessed the ability of DBSI, diffusion kurtosis imaging (DKI), generalized q‐sampling imaging (GQI), q‐ball imaging (QBI) and neurite orientation dispersion and density imaging to resolve fiber crossing, in reference to the gold standard angles measured from structural images. DTI‐computed diffusivities and fractional anisotropy were significantly confounded by gel‐mimicked edema and crossing fibers. Conversely, DBSI calculated accurate diffusivities of individual fibers regardless of the extent of simulated edema and degrees of fiber crossing angles. Additionally, DBSI accurately and consistently estimated crossing angles in various conditions of gel‐mimicked edema when compared with the gold standard (r² = 0.92, P = 1.9 × 10^?9, bias = 3.9°). Small crossing angles and edema significantly impact the diffusion orientation distribution function, making DKI, GQI and QBI less accurate in detecting and estimating fiber crossing angles. Lastly, we used diffusion tensor ellipsoids to demonstrate that DBSI resolves the confounds of edema and crossing fibers in the peritumoral edema region from a patient with lung cancer metastasis, while DTI failed. In summary, DBSI is able to separate two crossing fibers and accurately recover their diffusivities in a complex environment characterized by increasing crossing angles and amounts of gel‐mimicked edema. DBSI also indicated better angular resolution compared with DKI, QBI and GQI.     

Diffusion basis spectrum imaging detects and distinguishes coexisting subclinical inflammation,demyelination and axonal injury in experimental autoimmune encephalomyelitis mice

Clinicopathological paradox has hampered significantly the effective assessment of the efficacy of therapeutic intervention for multiple sclerosis. Neuroimaging biomarkers of tissue injury could guide more effective treatment by accurately reflecting the underlying subclinical pathologies. Diffusion tensor imaging‐derived directional diffusivity and anisotropy indices have been applied to characterize white matter disorders. However, these biomarkers are sometimes confounded by the complex pathologies seen in multiple sclerosis and its animal models. Recently, a novel technique of diffusion basis spectrum imaging has been developed to quantitatively assess axonal injury, demyelination and inflammation in a mouse model of inflammatory demyelination. Lenaldekar, which inhibits T‐cell expansion in a non‐cytolytic manner, has been shown to suppress relapses and preserve white matter integrity in mice with experimental autoimmune encephalomyelitis. In this study, relapsing–remitting experimental autoimmune encephalomyelitis was induced through active immunization of SJL/J mice with a myelin proteolipid protein peptide. The therapeutic efficacy of Lenaldekar treatment was evaluated via daily clinical score, cross‐sectional ex vivo diffusion basis spectrum imaging examination and histological analysis. Lenaldekar greatly reduced relapse severity and protected white matter integrity in these experimental autoimmune encephalomyelitis mice. Diffusion basis spectrum imaging‐derived axial diffusivity, radial diffusivity and restricted diffusion tensor fraction accurately reflected axonal injury, myelin integrity and inflammation‐associated cellularity change, respectively. These results support the potential use of diffusion basis spectrum imaging as an effective outcome measure for preclinical drug evaluation. Copyright © 2014 John Wiley & Sons, Ltd.     

Clinical evaluation of time‐of‐flight MR angiography with sparse undersampling and iterative reconstruction for cerebral aneurysms

Compressed sensing (CS) MRI has just been introduced to research areas as an innovative approach to accelerate MRI. CS is expected to achieve higher k‐space undersampling by exploiting the underlying sparsity in an appropriate transform domain. MR angiography (MRA) provides high spatial resolution information on arteries; however, a relatively long acquisition time is necessary to cover a wide volume. Reduction of acquisition time by CS for time‐of‐flight (TOF) MR angiography (Sparse‐TOF) is beneficial in clinical examinations; therefore, the clinical validity of Sparse‐TOF needs to be investigated. The aim of this study was to compare the diagnostic capability of TOF MRA between parallel imaging (PI)‐TOF with an acceleration factor of 3 (annotated as 3×) and Sparse‐TOF (3× and 5×) in patients with cerebral aneurysms. PI‐TOF (3×) and Sparse‐TOF (3× and 5×) imaging were performed in 20 patients using a 3 T MRI system. Aneurysms in PI‐TOF (3×) and Sparse‐TOF (3× and 5×) were blindly rated as visible or scarcely visible by neuroradiologists. The neck, height and width of aneurysms were also measured. Twenty‐six aneurysms were visualized and rated as visible in PI‐TOF (3×) and Sparse‐TOF (3× and 5×), with excellent agreement between two raters. No significant differences were found in measured neck, height or width of aneurysms among them. Sparse‐TOF (3× and 5×) were acquired and reconstructed within 6 min, and cerebral aneurysms were visible in both of them with equivalent quality to PI‐TOF (3×). Sparse‐TOF (5×) is a good alternative to PI‐TOF (3×) to visualize cerebral aneurysms.     

Fast and accurate initialization of the free‐water imaging model parameters from multi‐shell diffusion MRI

Cerebrospinal fluid partial volume effect is a known bias in the estimation of Diffusion Tensor Imaging (DTI) parameters from diffusion MRI data. The Free‐Water Imaging model for diffusion MRI data adds a second compartment to the DTI model, which explicitly accounts for the signal contribution of extracellular free‐water, such as cerebrospinal fluid. As a result the DTI parameters obtained through the free‐water model are corrected for partial volume effects, and thus better represent tissue microstructure. In addition, the model estimates the fractional volume of free‐water, and can be used to monitor changes in the extracellular space. Under certain assumptions, the model can be estimated from single‐shell diffusion MRI data. However, by using data from multi‐shell diffusion acquisitions, these assumptions can be relaxed, and the fit becomes more robust. Nevertheless, fitting the model to multi‐shell data requires high computational cost, with a non‐linear iterative minimization, which has to be initialized close enough to the global minimum to avoid local minima and to robustly estimate the model parameters. Here we investigate the properties of the main initialization approaches that are currently being used, and suggest new fast approaches to improve the initial estimates of the model parameters. We show that our proposed approaches provide a fast and accurate initial approximation of the model parameters, which is very close to the final solution. We demonstrate that the proposed initializations improve the final outcome of non‐linear model fitting.     

Partitioning k‐space for cylindrical three‐dimensional rapid acquisition with relaxation enhancement imaging in the mouse brain

Three‐dimensional rapid acquisition with relaxation enhancement (RARE) scans require the assignment of each phase encode step in two dimensions to an echo in the echo train. Although this assignment is frequently made across the entire Cartesian grid, collection of only the central cylinder of k‐space by eliminating the corners in each phase encode dimension reduces the scan time by ~22% with negligible impact on image quality. The recipe for the assignment of echoes to grid points for such an acquisition is less straightforward than for the simple full Cartesian acquisition case, and has important implications for image quality. We explored several methods of partitioning k‐space—exploiting angular symmetry in one extreme or emulating a cropped Cartesian acquisition in the other—and acquired three‐dimensional RARE magnetic resonance imaging (MRI) scans of the ex vivo mouse brain. We evaluated each partitioning method for sensitivity to artifacts and then further considered strategies to minimize these through averaging or interleaving of echoes and by empirical phase correction. All scans were collected 16 at a time with multiple‐mouse MRI. Although all schemes considered could be used to generate images, the results indicate that the emulation of a standard Cartesian echo assignment, by partitioning preferentially along one dimension within the cylinder, is more robust to artifacts. Samples at the periphery of the bore showed larger phase deviations and higher sensitivity to artifacts, but images of good quality could still be obtained with an optimized acquisition protocol. A protocol for high‐resolution (40 μm) ex vivo images using this approach is presented, and has been used routinely with a success rate of 99% in over 1000 images.     

Determination of the appropriate b value and number of gradient directions for high‐angular‐resolution diffusion‐weighted imaging

High‐angular‐resolution diffusion‐weighted imaging (HARDI) is one of the most common MRI acquisition schemes for use with higher order models of diffusion. However, the optimal b value and number of diffusion‐weighted (DW) directions for HARDI are still undetermined, primarily as a result of the large number of available reconstruction methods and corresponding parameters, making it impossible to identify a single criterion by which to assess performance. In this study, we estimate the minimum number of DW directions and optimal b values required for HARDI by focusing on the angular frequency content of the DW signal itself. The spherical harmonic (SH) series provides the spherical analogue of the Fourier series, and can hence be used to examine the angular frequency content of the DW signal. Using high‐quality data acquired along 500 directions over a range of b values, we estimate that SH terms above l = 8 are negligible in practice for b values up to 5000 s/mm², implying that a minimum of 45 DW directions is sufficient to fully characterise the DW signal. l > 0 SH terms were found to increase as a function of b value, levelling off at b = 3000 s/mm², suggesting that this value already provides the highest achievable angular resolution. In practice, it is recommended to acquire more than the minimum of 45 DW directions to avoid issues with imperfections in the uniformity of the DW gradient directions and to meet signal‐to‐noise requirements of the intended reconstruction method. Copyright © 2013 John Wiley & Sons, Ltd.     

Tract‐specific and age‐related variations of the spinal cord microstructure: a multi‐parametric MRI study using diffusion tensor imaging (DTI) and inhomogeneous magnetization transfer (ihMT)

Being able to finely characterize the spinal cord (SC) microstructure and its alterations is a key point when investigating neural damage mechanisms encountered in different central nervous system (CNS) pathologies, such as multiple sclerosis, amyotrophic lateral sclerosis or myelopathy. Based on novel methods, including inhomogeneous magnetization transfer (ihMT) and dedicated SC probabilistic atlas post‐processing, the present study focuses on the in vivo characterization of the healthy SC tissue in terms of regional microstructure differences between (i) upper and lower cervical vertebral levels and (ii) sensory and motor tracts, as well as differences attributed to normal aging. Forty‐eight healthy volunteers aged from 20 to 70 years old were included in the study and scanned at 3 T using axial high‐resolution T₂*‐w imaging, diffusion tensor imaging (DTI) and ihMT, at two vertebral levels (C2 and C5). A processing pipeline with minimal user intervention, SC segmentation and spatial normalization into a reference space was implemented in order to assess quantitative morphological and structural parameters (cross‐sectional areas, scalar DTI and MT/ihMT metrics) in specific white and gray matter regions of interest. The multi‐parametric MRI metrics collected allowed upper and lower cervical levels to be distinguished, with higher ihMT ratio (ihMTR), higher axial diffusivity (λ_∥) and lower radial diffusivity (λ_⊥) at C2 compared with C5. Significant differences were also observed between white matter fascicles, with higher ihMTR and lower λ_∥ in motor tracts compared with posterior sensory tracts. Finally, aging was found to be associated with significant metric alterations (decreased ihMTR and λ_∥). The methodology proposed here, which can be easily transferred to the clinic, provides new insights for SC characterization. It bears great potential to study focal and diffuse SC damage in neurodegenerative and demyelinating diseases. Copyright © 2016 John Wiley & Sons, Ltd.     

Comparison of image sensitivity between conventional tensor‐based and fast diffusion kurtosis imaging protocols in a rodent model of acute ischemic stroke

Diffusion kurtosis imaging (DKI) can offer a useful complementary tool to routine diffusion MRI for improved stratification of heterogeneous tissue damage in acute ischemic stroke. However, its relatively long imaging time has hampered its clinical application in the emergency setting. A recently proposed fast DKI approach substantially shortens the imaging time, which may help to overcome the scan time limitation. However, to date, the sensitivity of the fast DKI protocol for the imaging of acute stroke has not been fully described. In this study, we performed routine and fast DKI scans in a rodent model of acute stroke, and compared the sensitivity of diffusivity and kurtosis indices (i.e. axial, radial and mean) in depicting acute ischemic lesions. In addition, we analyzed the contrast‐to‐noise ratio (CNR) between the ipsilateral ischemic and contralateral normal regions using both conventional and fast DKI methods. We found that the mean kurtosis shows a relative change of 47.1 ± 7.3% between the ischemic and contralateral normal regions, being the most sensitive parameter in revealing acute ischemic injury. The two DKI methods yielded highly correlated diffusivity and kurtosis measures and lesion volumes (R² ? 0.90, p < 0.01). Importantly, the fast DKI method exhibited significantly higher CNR of mean kurtosis (1.6 ± 0.2) compared with the routine tensor protocol (1.3 ± 0.2, p < 0.05), with its CNR per unit time (CNR efficiency) approximately doubled when the scan time was taken into account. In conclusion, the fast DKI method provides excellent sensitivity and efficiency to image acute ischemic tissue damage, which is essential for image‐guided and individualized stroke treatment. Copyright © 2016 John Wiley & Sons, Ltd.     

The effects of ageing on mouse muscle microstructure: a comparative study of time‐dependent diffusion MRI and histological assessment

The investigation of age‐related changes in muscle microstructure between developmental and healthy adult mice may help us to understand the clinical features of early‐onset muscle diseases, such as Duchenne muscular dystrophy. We investigated the evolution of mouse hind‐limb muscle microstructure using diffusion imaging of in vivo and in vitro samples from both actively growing and mature mice. Mean apparent diffusion coefficients (ADCs) of the gastrocnemius and tibialis anterior muscles were determined as a function of diffusion time (Δ), age (7.5, 22 and 44 weeks) and diffusion gradient direction, applied parallel or transverse to the principal axis of the muscle fibres. We investigated a wide range of diffusion times with the goal of probing a range of diffusion lengths characteristic of muscle microstructure. We compared the diffusion time‐dependent ADC of hind‐limb muscles with histology. ADC was found to vary as a function of diffusion time in muscles at all stages of maturation. Muscle water diffusivity was higher in younger (7.5 weeks) than in adult (22 and 44 weeks) mice, whereas no differences were observed between the older ages. In vitro data showed the same diffusivity pattern as in vivo data. The highlighted differences in diffusion properties between young and mature muscles suggested differences in underlying muscle microstructure, which were confirmed by histological assessment. In particular, although diffusion was more restricted in older muscle, muscle fibre size increased significantly from young to adult age. The extracellular space decreased with age by only ~1%. This suggests that the observed diffusivity differences between young and adult muscles may be caused by increased membrane permeability in younger muscle associated with properties of the sarcolemma.     

Rapid reconstruction of highly undersampled,non‐Cartesian real‐time cine k‐space data using a perceptual complex neural network (PCNN)

Highly accelerated real‐time cine MRI using compressed sensing (CS) is a promising approach to achieve high spatio‐temporal resolution and clinically acceptable image quality in patients with arrhythmia and/or dyspnea. However, its lengthy image reconstruction time may hinder its clinical translation. The purpose of this study was to develop a neural network for reconstruction of non‐Cartesian real‐time cine MRI k‐space data faster (<1 min per slice with 80 frames) than graphics processing unit (GPU)‐accelerated CS reconstruction, without significant loss in image quality or accuracy in left ventricular (LV) functional parameters. We introduce a perceptual complex neural network (PCNN) that trains on complex‐valued MRI signal and incorporates a perceptual loss term to suppress incoherent image details. This PCNN was trained and tested with multi‐slice, multi‐phase, cine images from 40 patients (20 for training, 20 for testing), where the zero‐filled images were used as input and the corresponding CS reconstructed images were used as practical ground truth. The resulting images were compared using quantitative metrics (structural similarity index (SSIM) and normalized root mean square error (NRMSE)) and visual scores (conspicuity, temporal fidelity, artifacts, and noise scores), individually graded on a five‐point scale (1, worst; 3, acceptable; 5, best), and LV ejection fraction (LVEF). The mean processing time per slice with 80 frames for PCNN was 23.7 ± 1.9 s for pre‐processing (Step 1, same as CS) and 0.822 ± 0.004 s for dealiasing (Step 2, 166 times faster than CS). Our PCNN produced higher data fidelity metrics (SSIM = 0.88 ± 0.02, NRMSE = 0.014 ± 0.004) compared with CS. While all the visual scores were significantly different (P < 0.05), the median scores were all 4.0 or higher for both CS and PCNN. LVEFs measured from CS and PCNN were strongly correlated (R² = 0.92) and in good agreement (mean difference = ?1.4% [2.3% of mean]; limit of agreement = 10.6% [17.6% of mean]). The proposed PCNN is capable of rapid reconstruction (25 s per slice with 80 frames) of non‐Cartesian real‐time cine MRI k‐space data, without significant loss in image quality or accuracy in LV functional parameters.     

Fast low‐angle shot diffusion tensor imaging with stimulated echo encoding in the muscle of rabbit shank

In the past, spin‐echo (SE) echo planar imaging(EPI)‐based diffusion tensor imaging (DTI) has been widely used to study the fiber structure of skeletal muscles in vivo. However, this sequence has several shortcomings when measuring restricted diffusion in small animals, such as its sensitivity to susceptibility‐related distortions and a relatively short applicable diffusion time. To address these limitations, in the current work, a stimulated echo acquisition mode (STEAM) MRI technique, in combination with fast low‐angle shot (FLASH) readout (turbo‐STEAM MRI), was implemented and adjusted for DTI in skeletal muscles. Signal preparation using stimulated echoes enables longer effective diffusion times, and thus the detection of restricted diffusion within muscular tissue with intracellular distances up to 100 µm. Furthermore, it has a reduced penalty for fast T₂ muscle signal decay, but at the expense of 50% signal loss compared with a SE preparation. Turbo‐STEAM MRI facilitates high‐resolution DTI of skeletal muscle without introducing susceptibility‐related distortions. To demonstrate its applicability, we carried out rabbit in vivo measurements on a human whole‐body 3 T scanner. DTI parameters of the shank muscles were extracted, including the apparent diffusion coefficient, fractional anisotropy, eigenvalues and eigenvectors. Eigenvectors were used to calculate maps of structural parameters, such as the planar index and the polar coordinates θ and ? of the largest eigenvector. These parameters were compared between three muscles. θ and ? showed clear differences between the three muscles, reflecting different pennation angles of the underlying fiber structures. Fiber tractography was performed to visualize and analyze the architecture of skeletal pennate muscles. Optimization of tracking parameters and utilization of T₂‐weighted images for improved muscle boundary detection enabled the determination of additional parameters, such as the mean fiber length. The presented results support the applicability of turbo‐STEAM MRI as a promising method for quantitative DTI analysis and fiber tractography in skeletal muscles. Copyright © 2013 John Wiley & Sons, Ltd.