Fast MR Research Articles

Standard Normal Fetal Lung Volume by MRI Measurement

Background: Pulmonary hypoplasia is a significant decrease in lung volume by any cause in comparison to the normal volume for gestational age. Objectives: To evaluate the application of prenatal magnetic resonance imaging in order to predict pulmonary hypoplasia, measurement of fetal lung volume (FLV) was aimed to establish a reference data in normal fetuses. Patients and Methods: Three hundred forty two fetuses with abnormal ultrasound findings underwent fast spin-echo T2-weighted MR imaging at a single tertiary care hospital (Shahid Faghihi Hospital, Shiraz, Iran). Data of 241 fetuses at 18 - 36 weeks’ gestation without thoracic malformations were obtained for a FLV normative curve. To acquire a best-fit formula for the mean total fetal lung volume based on each gestational week with 95% confidence interval (CI), we used the regression model applied by Rypens et al. This resulted in a formula calculating the expected fetal lung volume based on gestational age. A one-sample t-test was also carried out to compare the mean total fetal lung volume at each gestational week with the expected mean total fetal lung volume predicted by the formula by Rypens et al., Meyers et al., and Osada et al. Results: Normal FLV increased with gestational age as the spread of values. The expected fetal lung volume was derived from the following formula: Fetal lung volume (mL) = 0.002 × (gestational age [GA]2.913), in which GA is gestational weeks. FLV in our population had a more consistent correlation with Japanese fetuses than European or American fetuses. Conclusion: In fetuses with normal lungs, FLV distribution against gestational age is simply measured prenatally with MR imaging. A single universal formula might not be suitable for fetuses worldwide. In our region, acquired data were more consistent with Asian normal values.

2D multislice MP2RAGE sequence for fast T1 mapping at 7 T: Application to mouse imaging and MR thermometry.

To develop a 2D radial multislice MP2RAGE sequence for fast and reliable T1 mapping at 7 T in mice and for MR thermometry. The 2D-MP2RAGE sequence was performed with the following parameters: TI1 -TI2 -MP2RAGETR = 1000-3000-9000 ms. The multiple dead times within the sequence were used for interleaved multislice acquisition, enabling one to acquire six slices in 9 seconds. The excitation pulse shape, inversion selectivity, and interslice gap were optimized. In vitro comparison with the inversion-recovery sequence was performed. The T1 variations with temperature were measured on tubes with T1 ranging from 800 ms to 2000 ms. The sequence was used to acquire T1 maps continuously during 30 minutes on the brain and abdomen of healthy mice. A three-lobe cardinal sine excitation pulse, combined with an inversion slice thickness and an interslice gap of respectively 150% and 50% of the imaging slice thickness, led to a SD and bias of the T1 measurements below 1% and 2%, respectively. A linear dependence of T1 with temperature was measured between 10°C and 60°C. In vivo, less than 1% variation was measured between successive T1 maps in the mouse brain. In the abdomen, no obvious in-plane motion artifacts were observed but respiratory motion in the slice dimension led to 6% T1 underestimation. The multislice MP2RAGE sequence could be used for fast whole-body T1 mapping and MR thermometry. Its reconstruction method would enable on-the-fly reconstruction.

DeepcomplexMRI: Exploiting deep residual network for fast parallel MR imaging with complex convolution

This paper proposes a multi-channel image reconstruction method, named DeepcomplexMRI, to accelerate parallel MR imaging with residual complex convolutional neural network. Different from most existing works which rely on the utilization of the coil sensitivities or prior information of predefined transforms, DeepcomplexMRI takes advantage of the availability of a large number of existing multi-channel groudtruth images and uses them as target data to train the deep residual convolutional neural network offline. In particular, a complex convolutional network is proposed to take into account the correlation between the real and imaginary parts of MR images. In addition, the k-space data consistency is further enforced repeatedly in between layers of the network. The evaluations on in vivo datasets show that the proposed method has the capability to recover the desired multi-channel images. Its comparison with state-of-the-art methods also demonstrates that the proposed method can reconstruct the desired MR images more accurately.

Usefulness of iodine-blood material density images in estimating degree of liver fibrosis by calculating extracellular volume fraction obtained from routine dual-energy liver CT protocol equilibrium phase data: preliminary experience.

To assess whether extracellular volume fraction (ECV) calculated from iodine(-blood) density images (I-B) of dual-energy liver CT (DECT) equilibrium phase data (EqD) is useful in estimating the degree of liver fibrosis. Consecutive 52 patients with chronic liver disease who underwent fast kV switching DECT and liver MR elastography (MRE) were retrospectively enrolled. Iodine(-water) density images (I-W) and I-B generated from EqD and ECV were calculated. As blood pools, abdominal aorta (Ao) and suprahepatic inferior vena cava (IVC) were chosen, and, therefore, 4 types of ECV (ECVI-W Ao, ECVI-W IVC, ECVI-B Ao, ECVI-B IVC) were obtained. ECV was also calculated using conventional method (ECVconv Ao). The correlation coefficients (R2 or rho) of these five ECVs versus liver stiffness (MRE) or pathologically proven fibrosis grades were compared. As for correlation with liver stiffness, R2 for ECVconv.Ao, ECVI-W Ao, ECVI-B Ao, ECVI-W IVC, and ECVI-B IVC, were 0.26, 0.34, 0.44, 0.39, and 0.52, respectively (all p < 0.0001). Histopathological correlation was available in 28 patients, and rho values were 0.61, 0.60, 0.71, 0.68, and 0.76, respectively (all p < 0.001). ECVI-B IVC calculated from EqD of DECT is useful in estimating the degree of liver fibrosis.

Measuring Tumor Glycolytic Flux in Vivo by Using Fast Deuterium MRI.

Background Tumor cells frequently show high rates of aerobic glycolysis, which provides the glycolytic intermediates needed for the increased biosynthetic demands of rapid cell growth and proliferation. Existing clinical methods (fluorodeoxyglucose PET and carbon 13 MRI and spectroscopy) do not allow quantitative images of glycolytic flux. Purpose To evaluate the use of deuterium (hydrogen 2 [2H]) MR spectroscopic imaging for quantitative mapping of tumor glycolytic flux and to assess response to chemotherapy. Materials and Methods A fast three-dimensional 2H MR spectroscopic imaging pulse sequence, with a time resolution of 10 minutes, was used to image glycolytic flux in a murine tumor model after bolus injection of D-[6,6'-2H2]glucose before and 48 hours after treatment with a chemotherapeutic agent. Tumor lactate labeling, expressed as the lactate-to-water and lactate-to-glucose signal ratios, was also assessed in localized 2H MR spectra. Statistical significance was tested with a one-sided paired t test. Results 2H MR spectroscopic imaging showed heterogeneity in glycolytic flux across the tumor and an early decrease in flux following treatment with a chemotherapeutic drug. Spectroscopy measurements on five animals showed a decrease in the lactate-to-water signal ratio, from 0.33 ± 0.10 to 0.089 ± 0.039 (P = .005), and in the lactate-to-glucose ratio, from 0.27 ± 0.12 to 0.12 ± 0.06 (P = .04), following drug treatment. Conclusion Rapidly acquired deuterium (hydrogen 2) MR spectroscopic images can provide quantitative and spatially resolved measurements of glycolytic flux in tumors that can be used to assess treatment response. Published under a CC BY 4.0 license. Online supplemental material is available for this article. See also the editorial by Ouwerkerk in this issue.

A flexible 9-channel coil array for fast 3D MR thermometry in MR-guided high-intensity focused ultrasound (HIFU) studies on rabbits at 3 T

Signal-to-noise ratio (SNR) is a critical factor in MR-guided high-intensity focused ultrasound (HIFU) for local heating, which can affect the accuracy of temperature measurement. In order to achieve high SNR and higher temporal resolution, dedicated coil arrays for MR-guided HIFU applications need to be developed. In this work, a flexible 9-channel coil array was designed, and constructed at 3 T to achieve fast temperature mapping for MR-guided HIFU applications on rabbit leg muscle. Coil performance was evaluated for SNR, and parallel imaging capability by in-vivo studies. Compared to a commercially available 4-channel flexible coil array, the dedicated 9-channel coil array has a much higher SNR, with at least a 2.6-fold increment in the region of interest (ROI). The inverse g-factors maps demonstrated that the dedicated 9-channel coil array has a better parallel imaging capability than the Flex Small 4. With accelerations normal to the array direction, both coil arrays showed much higher g-factors than those of accelerations along the array direction. Room temperature mapping was implemented to evaluate the temperature measurement accuracy by in-vivo experiments. The precisions of the 9-channel coil, ±0.18 °C for un-acceleration and ± 0.56 °C for acceleration at R = 2 × 2, both improved by an order of magnitude than these of the 4-channel coil, which were ± 1.45 °C for un-acceleration and ± 3.52 °C for acceleration at R = 2 × 2. In the fast temperature imaging on the rabbit leg muscle with heating, a high temporal resolution of 3.3 s with a temperature measurement precision of ±0.56 °C has been achieved using the dedicated 9-channel coil. This study demonstrates that the dedicated 9-channel coil array for rabbit leg imaging provides improved performance in SNR, parallel imaging capability, and the accuracy of temperature measurement compared to a commercial 4-channel coil, and it also achieves fast temperature mapping in practical MR-guided HIFU applications.

4324Fast-field cycling magnetic resonance detection of intracellular iron in the nanomolar range - A pre-requisite for in-vivo study of inflammation

Abstract Background Ultra-small superparamagnetic iron oxide (USPIO) nanoparticles are phagocytosed by macrophages and when subjected to magnetic resonance imaging (MRI) detect inflammation. We describe a novel MR technique where the magnetic field is rapidly cycled (Fast field-cycling MR, FFC-MR), assessing T1 dispersion over a range of low (100μT-0.2T) fields and offering superior T1-based iron quantification. AIM To quantify iron using the FFC-MR R1 (1/T1) dispersion profile and flow cytometry features of USPIO-laden cells, in comparison to colorimetric assays. Methods Murine J774 macrophage-like cells were incubated with 0–200μg/ml Fe as USPIO (ferumoxytol) for 16 hours, washed and suspended in 500μl PBS/2mM EDTA. Prussian blue staining confirmed USPIO phagocytosis. Relaxation was measured using a clinical, in-house built prototype 0.2T FFC-MR system with a custom test tube coil. R1 dispersion was derived from a saturation recovery sequence (Fig. 1A). R1 (s–1) values were calculated with a monoexponential curve fitting algorithm, R2 of fits were ≥0.999. R1 dispersion profiles were generated plotting R1 against the magnetic field (T). Quantification of side scatter (SCC) intensity and the USPIO-occupied fraction of total cell area was performed with imaging flow cytometry. A colorimetric assay provided validation of cell iron content. All data are mean±SEM, analysed with t-tests, Pearson correlation and linear regression; statistical significance set at p&lt;0.05. Results Table 1 shows quantitative data derived by all 3 modalities with increasing USPIO exposure. FFC-MR detection of intracellular iron was excellent (p≤0.001 vs. control for all), with separation of average R1 dispersion profiles (Fig. 1B), strong correlation with colorimetry (r=0.993 p&lt;0.001) and good fit on linear regression model (R2=0.9222, Fig 1C). Flow cytometry quantification of SCC was comparable (p≤0.001 for all), whereas USPIO-occupied area was only sensitive at exposures ≥10μg/ml USPIO (r=0.967 p&lt;0.001 &amp; r=0.983 p&lt;0.001 vs. colorimetric respectively). FFC-MR detected iron at ≥1.12ng/μg protein. USPIO exposure (μg/ml medium) 0 5 10 40 80 100 200 FFC-MR Mean R1 1x106 cell suspension (s–1) 0.308±0.014 0.356±0.013** 0.432±0.016** 0.706±0.021** 1.174±0.031** 1.239±0.033** 1.599±0.041** Flow cytometry USPIO area/cell area 0.034±0.001 0.036±0.001 0.037±0.001* 0.069±0.001** 0.085±0.001** 0.090±0.001** 0.097±0.001** Flow cytometry SSC intensity 26860 32815** 39573** 69285** 80967** 82693** 86373** Colorimetric assay Iron concentration 1x106 cells (ng/μg protein) 0.115±0.118 1.121±0.045** 2.074±0.084** 5.496±0.134** 8.421±0.269** 9.771±0.100** 12.398±0.233** SSC = side-scattered light; *p&lt;0.05 vs. control; **p&lt;0.001 vs. control. Conclusion Field-cycling MR is capable of highly accurate intracellular USPIO quantification, which has potential to non-invasively detect clinically relevant amounts of iron in inflammatory cardiovascular diseases. Acknowledgement/Funding NHS Grampian Endowment Fund

MR fingerprinting for water T1 and fat fraction quantification in fat infiltrated skeletal muscles.

To develop a fast MR fingerprinting (MRF) sequence for simultaneous estimation of water T1 (T1H2O ) and fat fraction (FF) in fat infiltrated skeletal muscles. The MRF sequence for T1H2O and FF quantification (MRF T1-FF) comprises a 1400 radial spokes echo train, following nonselective inversion, with varying echo and repetition time, as well as prescribed flip angle. Undersampled frames were reconstructed at different acquisition time-points by nonuniform Fourier transform, and a bi-component model based on Bloch simulations applied to adjust the signal evolution and extract T1H2O and FF. The sequence was validated on a multi-vial phantom, in three healthy volunteers and five patients with neuromuscular diseases. We evaluated the agreement between MRF T1-FF parameters and reference values and confounding effects due to B0 and B1 inhomogeneities. In phantom, T1H2O and FF were highly correlated with references values measured with multi-inversion time inversion recovery-stimulated echo acquisition mode and Dixon, respectively (R2 > 0.99). In vivo, T1H2O and FF determined by the MRF T1-FF sequence were also correlated with reference values (R2 = 0.98 and 0.97, respectively). The precision on T1H2O was better than 5% for muscles where FF was less than 0.4. Both T1H2O and FF values were not confounded by B0 nor B1 inhomogeneities. The MRF T1-FF sequence derived T1H2O and FF values in voxels containing a mixture of water and fat protons. This method can be used to comprehend and characterize the effects of tissue water compartmentation and distribution on muscle T1 values in patients affected by chronic fat infiltration.

Desulfonation of the textile azo dye Acid Fast Yellow MR by newly isolated Aeromonas hydrophila SK16

Acid Fast Yellow MR (AFYMR), a sulphonated reactive textile azo dye, was effectively decolorized by a newly identified bacterial strain Aeromonas hydrophila SK16. Nineteen isolates were obtained from the textile effluent polluted soil in Perundurai region. Maximum decolorization of AFYMR was observed in Aeromonas hydrophila SK16. Enzyme activities present in A. hydrophila SK16 revealed the involvement of oxidoreductase and efficiently decolorized 91.25% of AFYMR (100 mg L-l) in static condition for 9 h. Fourier Transform Infrared Spectroscopy (FTIR) analysis showed that A. hydrophila SK16 cells preferred N–H and Sulphur dioxide bonds in the breakdown of AFYMR. The final metabolite supports the degradation of AFYMR through desulfonation as revealed by Gas Chromatography-Mass Spectroscopy (GCMS) results. Phytotoxicity studies conducted in Oryza sativa and Zea mays illustrated significant decrease in toxicity of metabolite produced during decolorization of Acid Fast Yellow MR by Aeromonas hydrophila SK16.

Highly undersampled magnetic resonance imaging reconstruction using autoencoding priors.

Although recent deep learning methodologies have shown promising results in fast MR imaging, how to explore it to learn an explicit prior and leverage it into the observation constraint is still desired. A denoising autoencoder (DAE) network is leveraged as an explicit prior to address the highly undersampling MR image reconstruction problem. First, inspired by the observation that the prior information learned from high-dimension signals is more effective than that from the low-dimension counterpart in image restoration tasks, we train the network in a multichannel scenario and apply the learned network to single-channel image reconstruction by a variables augmentation technique. Second, because of the fact that multiple implementations of artificial noise generation in DAE favors a better underlying result, we introduce a 2-sigma rule to complement each other for improving the final reconstruction. The whole algorithm is tackled by proximal gradient descent. Experimental results under varying sampling trajectories and acceleration factors consistently demonstrate the superiority of the enhanced autoencoding priors, in terms of peak signal-to-noise ratio, structural similarity, and high-frequency error norm. A simple and effective way to incorporate the DAE prior into highly undersampling MR reconstruction is proposed. Once the DAE prior is obtained, it can be applied to the reconstruction tasks with different sampling trajectories and acceleration factors, and achieves superior performance in comparison with state-of-the-art methods.

Proton MR spectroscopy and the detection of malignancy in ovarian masses.

To assess the impact of MR spectroscopy (MRS) on the detection of malignancy in ovarian masses. This prospective work included 230 females that had 245 adnexal/ovarian masses. Tumours were spotted by preliminary pelvic ultrasound. Masses assessed by MRI, multi- or single-voxel spectroscopy. Patients' spectra were assessed for peaks of lactate (Lac, 1.31 ppm), lipid (Lip, 1.33 ppm), N-acetyl aspartate (2.0 ppm), acetone (A, 2.05 ppm), choline (Cho, 3.23 ppm) and creatinine (Cr, 3.4 ppm) and the mean values of the (Cho/Cr) ratios were performed by a semi-quantitative approach. The operative pathology served as the standard of reference. Cho peak twofold higher than the average noise level was detected in 72% of the malignant and only 5.4% of the benign masses with an accuracy of 83%. Adding lactate to the choline enhanced the accuracy to 93%. The mean Cho/Cr ratios of the malignant ovarian masses (2.8) were significantly higher than that of the benign ones (1.2) . We used a receiver operating characteristic curve to determine the best cut-off value (1.7) for the mean Cho/Cr ratio to discriminate malignancy with sensitivity: 81.2%, specificity: 93.3 %, positive-predictive value: 92.9 %, negative-predictive value: 82.4% and accuracy: 87.1%. The simultaneous presence of choline and lactate peaks in MRS examination of the ovarian masses minimizes the overlap between benign and malignant categories. N-acetyl aspartate and acetone are the metabolites for diagnosing complex cystic masses as benign teratoma, endomterioma and tubo- ovarian abscess. MRS is a non-contrast based and fast MR sequence that gives an idea about tissue components could be used to improve the sensitivity and the accuracy of detecting malignancy in ovarian masses.

Three Dimensional Positive Contrast Susceptibility Fast Spin Echo MR Imaging with Variable Excitation Pulses and PD Algorithm.

A susceptibility-based positive contrast MR technique is applied to image the MR compatible metallic devices by solving a regularized ℓ1 minimization problem. However, the previous SE/FSE sequence is used for the data acquisition which can result in high SAR and low sampling efficiency in 3D imaging. Therefore, a 3D single slab 3D FSE sequence with slab selective and variable excitation pulse is proposed to implement 3D positive contrast MR imaging for low SAR and acquiring high-resolution 3D images within a shorter timeframe. Furthermore, in order to achieve faster reconstruction and better imaging quality of the 3D positive contrast MRI, the primal-dual iteration algorithm is also used to solve the regularized ℓ1 minimization problem. The visualization of the positive contrast and convergence behaviour of the proposed reconstruction framework base on the first-order PD algorithm were tested and validated on phantom experiments, compared with the previous nonlinear conjugate gradient (NLCG), fast iterative soft thresholding (FISTA) and alternating direction method of multipliers (ADMM) algorithms.

Parallel imaging and convolutional neural network combined fast MR image reconstruction: Applications in low-latency accelerated real-time imaging.

To develop and evaluate a parallel imaging and convolutional neural network combined image reconstruction framework for low-latency and high-quality accelerated real-time MR imaging. Conventional Parallel Imaging reconstruction resolved as gradient descent steps was compacted as network layers and interleaved with convolutional layers in a general convolutional neural network. All parameters of the network were determined during the offline training process, and applied to unseen data once learned. The proposed network was first evaluated for real-time cardiac imaging at 1.5T and real-time abdominal imaging at 0.35T, using threefold to fivefold retrospective undersampling for cardiac imaging and threefold retrospective undersampling for abdominal imaging. Then, prospective undersampling with fourfold acceleration was performed on cardiac imaging to compare the proposed method with standard clinically available GRAPPA method and the state-of-the-art L1-ESPIRiT method. Both retrospective and prospective evaluations confirmed that the proposed network was able to images with a lower noise level and reduced aliasing artifacts in comparison with the single-coil based and L1-ESPIRiT reconstructions for cardiac imaging at 1.5T, and the GRAPPA and L1-ESPIRiT reconstructions for abdominal imaging at 0.35T. Using the proposed method, each frame can be reconstructed in less than 100ms, suggesting its clinical compatibility. The proposed Parallel Imaging and convolutional neural network combined reconstruction framework is a promising technique that allows low-latency and high-quality real-time MR imaging.

Fully Automated Diagnosis of Anterior Cruciate Ligament Tears on Knee MR Images by Using Deep Learning.

To investigate the feasibility of using a deep learning-based approach to detect an anterior cruciate ligament (ACL) tear within the knee joint at MRI by using arthroscopy as the reference standard. A fully automated deep learning-based diagnosis system was developed by using two deep convolutional neural networks (CNNs) to isolate the ACL on MR images followed by a classification CNN to detect structural abnormalities within the isolated ligament. With institutional review board approval, sagittal proton density-weighted and fat-suppressed T2-weighted fast spin-echo MR images of the knee in 175 subjects with a full-thickness ACL tear (98 male subjects and 77 female subjects; average age, 27.5 years) and 175 subjects with an intact ACL (100 male subjects and 75 female subjects; average age, 39.4 years) were retrospectively analyzed by using the deep learning approach. Sensitivity and specificity of the ACL tear detection system and five clinical radiologists for detecting an ACL tear were determined by using arthroscopic results as the reference standard. Receiver operating characteristic (ROC) analysis and two-sided exact binomial tests were used to further assess diagnostic performance. The sensitivity and specificity of the ACL tear detection system at the optimal threshold were 0.96 and 0.96, respectively. In comparison, the sensitivity of the clinical radiologists ranged between 0.96 and 0.98, while the specificity ranged between 0.90 and 0.98. There was no statistically significant difference in diagnostic performance between the ACL tear detection system and clinical radiologists at P < .05. The area under the ROC curve for the ACL tear detection system was 0.98, indicating high overall diagnostic accuracy. There was no significant difference between the diagnostic performance of the ACL tear detection system and clinical radiologists for determining the presence or absence of an ACL tear at MRI.© RSNA, 2019Supplemental material is available for this article.

Fast 3D brain MR fingerprinting based on multi-axis spiral projection trajectory.

To develop a fast, sub-millimeter 3D magnetic resonance fingerprinting (MRF) technique for whole-brain quantitative scans. An acquisition trajectory based on multi-axis spiral projection imaging (maSPI) was implemented for 3D MRF with steady-state precession and slab excitation. By appropriately assigning the in-plane and through-plane rotations of spiral interleaves in a novel acquisition scheme, an maSPI-based acquisition was implemented, and the total acquisition time was reduced by up to a factor of 8 compared to stack-of-spiral (SOS)-based acquisition. A sliding-window method was also used to further reduce the required number of time points for a faster acquisition. The experiments were conducted both on a phantom and in vivo. The results from the phantom measurements with the proposed and gold standard methods were consistent with a good linear correlation and an R2 value approaching 0.99. The in vivo experiments achieved whole-brain parametric maps with isotropic resolutions of 1 mm and 0.8 mm in 5.0 and 6.0 min, respectively, with potential for further acceleration. An in vivo experiment with intentionally moving subjects demonstrated that the maSPI scheme largely outperforms the SOS scheme in terms of robustness to head motion. 3D MRF with an maSPI acquisition scheme enables fast and robust scans for high-resolution parametric mapping.

Case 267.

History A 28-year-old man presented with lifelong anejaculation, which had become an issue because of family planning. The patient had a history of normal erections and experienced the sensation of orgasm without ever ejaculating. On physical examination, both testes were present in the scrotum, with normal dimensions and a normal epididymis bilaterally. The patient had a slightly tender left testicle, and digital rectal examination findings were normal. The patient underwent further investigation for the possibility of retrograde ejaculation with urine cytology, the results of which were negative. Genetic testing was performed to exclude Y chromosome microdeletions. Serum-luteinizing and follicle-stimulating hormone levels were normal, with a borderline low level of testosterone (7.6 nmol/L; normal range, 8.0-29.0 nmol/L). All other pertinent laboratory results were noncontributory. Pelvic MRI was requested to exclude an anatomic cause of anejaculation. MRI was performed in accordance with the standard clinical prostate protocol, with a dynamic contrast material-enhanced study ( Figs 1 - 3 ). CT of the upper abdomen was also performed ( Fig 4 ). The patient subsequently underwent cystoscopy, which revealed an intravesicular fluid-filled mass near the left ureteric orifice ( Fig 5 ). Figure 1a: (a) Coronal and (b, c) axial fast spin-echo T2-weighted MR images of the pelvis, with b being superior to c. Figure 1b: (a) Coronal and (b, c) axial fast spin-echo T2-weighted MR images of the pelvis, with b being superior to c. Figure 1c: (a) Coronal and (b, c) axial fast spin-echo T2-weighted MR images of the pelvis, with b being superior to c. Figure 2a: (a) Coronal T2-weighted (repetition time msec/echo time msec, 4574/86.5) MR image of the pelvis. (b) Axial T2-weighted (3000/85.4) MR image of the pelvis. Figure 2b: (a) Coronal T2-weighted (repetition time msec/echo time msec, 4574/86.5) MR image of the pelvis. (b) Axial T2-weighted (3000/85.4) MR image of the pelvis. Figure 3: Unenhanced axial fat-saturated T1-weighted (6.2/3.1) MR images. Figure 4: Coronal CT urogram. Figure 5: Image obtained at cystoscopy.

The Many Uses of Chemical Shift Imaging in Body MRI

Chemical shift imaging is a relatively fast MR sequence, which is valuable in characterizing pathologies in multiple organ systems to help narrow the differential diagnosis or make a definitive diagnosis. Characterization by chemical shift imaging is based on three principles: detection of intracytoplasmic lipid, India-ink artifact, and susceptibility artifact.

Sensitivity analysis of neurodynamic and electromagnetic simulation parameters for robust prediction of peripheral nerve stimulation

Peripheral nerve stimulation (PNS) has become an important limitation for fast MR imaging using the latest gradient hardware. We have recently developed a simulation framework to predict PNS thresholds and stimulation locations in the body for arbitrary coil geometries to inform the gradient coil optimization process. Our approach couples electromagnetic field simulations in realistic body models to a neurodynamic model of peripheral nerve fibers. In this work, we systematically analyze the impact of key parameters on the predicted PNS thresholds to assess the robustness of the simulation results. We analyze the sensitivity of the simulated thresholds to variations of the most important simulation parameters, including parameters of the electromagnetic field simulations (dielectric tissue properties, body model size, position, spatial resolution, and coil model discretization) and parameters of the neurodynamic simulation (length of the simulated nerves, position of the nerve model relative to the extracellular potential, temporal resolution of the nerve membrane dynamics). We found that for the investigated setup, the subject-dependent parameters (e.g. tissue properties or body size) can affect PNS prediction by up to ~26% when varied in a natural range. This is in accordance with the standard deviation of ~30% reported in human subject studies. Parameters related to numerical aspects can cause significant simulation errors (>30%), if not chosen cautiously. However, these perturbations can be controlled to yield errors below 5% for all investigated parameters without an excessive increase in computation time. Our sensitivity analysis shows that patient-specific parameter fluctuations yield PNS threshold variations similar to the variations observed in experimental PNS studies. This may become useful to estimate population-average PNS thresholds and understand their standard deviation. Our analysis indicates that the simulated PNS thresholds are numerically robust, which is important for ranking different MRI gradient coil designs or assessing different PNS mitigation strategies.

ADMM-CSNet: A Deep Learning Approach for Image Compressive Sensing.

Compressive sensing (CS) is an effective technique for reconstructing image from a small amount of sampled data. It has been widely applied in medical imaging, remote sensing, image compression, etc. In this paper, we propose two versions of a novel deep learning architecture, dubbed as ADMM-CSNet, by combining the traditional model-based CS method and data-driven deep learning method for image reconstruction from sparsely sampled measurements. We first consider a generalized CS model for image reconstruction with undetermined regularizations in undetermined transform domains, and then two efficient solvers using Alternating Direction Method of Multipliers (ADMM) algorithm for optimizing the model are proposed. We further unroll and generalize the ADMM algorithm to be two deep architectures, in which all parameters of the CS model and the ADMM algorithm are discriminatively learned by end-to-end training. For both applications of fast CS complex-valued MR imaging and CS imaging of real-valued natural images, the proposed ADMM-CSNet achieved favorable reconstruction accuracy in fast computational speed compared with the traditional and the other deep learning methods.

Shearlet-based compressed sensing for fast 3D cardiac MR imaging using iterative reweighting

High-resolution three-dimensional (3D) cardiovascular magnetic resonance (CMR) is a valuable medical imaging technique, but its widespread application in clinical practice is hampered by long acquisition times. Here we present a novel compressed sensing (CS) reconstruction approach using shearlets as a sparsifying transform allowing for fast 3D CMR (3DShearCS) using 3D radial phase encoding (RPE). An iterative reweighting scheme was applied during image reconstruction to ensure fast convergence and high image quality. Shearlets are mathematically optimal for a simplified model of natural images and have been proven to be more efficient than classical systems such as wavelets. 3DShearCS was compared to three other commonly used reconstruction approaches. Image quality was assessed quantitatively using general image quality metrics and using clinical diagnostic scores from expert reviewers. The proposed technique had lower relative errors, higher structural similarity and higher diagnostic scores compared to the other reconstruction techniques especially for high undersampling factors, i.e. short scan times. 3DShearCS provided ensured accurate depiction of cardiac anatomy for fast imaging and could help to promote 3D high-resolution CMR in clinical practice.