NMR in Biomedicine最新文献

Muscle strength declines with aging at a faster rate compared with muscle mass, suggesting that not only muscle quantity but also muscle quality and architecture are age-dependent. This study tested the hypothesis that quantitative MRI (qMRI)-derived biomarkers of muscle quality (fractional anisotropy [FA], radial diffusivity [RD], axial diffusivity [AD], fat fraction [FF], and T₂ relaxation time) and architecture (fascicle length) could improve the prediction of skeletal muscle strength over muscle mass alone. We recruited 24 adults (12 female, age range 30-79 years). Muscle mass was estimated as the volume and cross-sectional area (CSA) of the quadriceps. FA, RD, and AD parameters, together with fascicle length for the rectus femoris (RF) and vastus lateralis (VL), were derived from diffusion tensor imaging (DTI), and muscle-T₂ was calculated from a multi-echo spin echo sequence. FF was determined using the Dixon approach. CSA values were combined with FF to calculate the lean CSA. Isometric, eccentric, and concentric knee extension torques were measured for the left and right leg using an isokinetic dynamometer. The univariable assessment of torque was performed using a linear regression. The statistical significance of adding qMRI parameters to the torque prediction models was tested using a mixed-effect regression. The best univariable predictor of isometric, eccentric, and concentric torque was lean CSA. Adding FA, RF fascicle length, and VL fascicle length to the model improved the prediction of concentric torque compared with CSA alone. The addition of FA, T₂, RD, RF fascicle length, and VL fascicle length improved the prediction of eccentric torque over CSA alone. The addition of FF was not significant within the model. Our results confirmed the hypothesis that the inclusion of qMRI parameters of muscle composition and architecture leads to higher R² coefficients for the prediction of muscle strength compared with models solely based on muscle quantity. These observations support the utility of qMRI for future research on sarcopenia prediction and management.

Diffusion-weighted imaging (DWI) during MR enterography helps identify bowel inflammation in Crohn's disease (CD). However, image quality is compromised by geometric distortions from B₀ field variations and physiological motion, making it challenging for radiologists to correlate findings between DWI and structural images. Traditional correction methods using reversed polarity scans are ineffective due to motion between acquisitions, which limits accurate estimation of intravoxel incoherent motion (IVIM) parameters. We propose a dual-echo echo-planar imaging (EPI) method that retrospectively corrects both geometric distortions and motion in 3T bowel DWI by accounting for field changes during peristalsis and breathing. We added a 5- to 7-min dual-echo EPI DW sequence (eight b-values, six directions) to the clinical MR enterography protocol of 21 patients with suspected CD at 3T MRI. Distortion correction was applied based on dynamically estimated fields from dual-echo DWI, followed by intra-volume registration between odd-even slices and inter-volume registration for motion correction. Two experienced board-certified radiologists evaluated the severity of the disease using simplified magnetic resonance index of activity (MaRIA) scores. Based on their consensus scores, patients were categorized into three groups: no active disease (MaRIA score = 0), active disease (MaRIA score = 1-2), and severe disease (MaRIA score = 3-5). The proposed DWI correction pipeline improved DWI/T₂-weighted image Dice similarity from 0.73 to 0.89, enabling better correlation of findings between structural and DW-MR images and enhancing DWI's clinical value. Corrected IVIM parameters showed stronger correlations with MaRIA scores (D: ρ = -0.93; f: ρ = -0.94, p < 0.001) compared to uncorrected parameters (D: ρ = -0.68, p = 0.001; f: ρ = -0.35, p = 0.118). Diagnostic sensitivity increased from 0.44 to 0.89, while parameter uncertainty decreased from 35.58% to 19.31% for D and 63.48% to 40.40% for f (p < 0.001). These improvements strengthen quantitative IVIM imaging for CD assessment, potentially reducing reliance on contrast imaging while offering enhanced tissue perfusion and diffusion insights.

Multi-component T1ρ mapping of the knee joint using nonlinear least squares (NLS)-based methods is usually a computationally intensive task, limiting its application to only a few voxels in the knee joint. Deep learning (DL) is a computationally fast alternative, but requires a large amount of training data. We propose the Synthetic data-Guided supervised DL Network (SGDNet) that utilizes synthetically generated data for training, eliminating the need for large datasets of T1ρ maps. Initially, residual connections are added to improve gradient flow and stabilize training. A self-attention module is also integrated into the SGDNet to obtain more accurate estimated relaxation maps. Additionally, to ensure both parameter fidelity and data consistency, we employ a customized loss function that penalizes discrepancies between actual and predicted T1ρ values as well as between measured and simulated MR signals. To combine speed and precision, we further introduce HSGDNet, a hybrid approach that uses SGDNet's outputs as initialization for a few NLS iterations. Extensive experimental analysis reveals that HSGDNet outperforms the competing methods by achieving average error reductions of 91.4%, 31.5%, and 36.0% for mono-exponential (ME), stretched-exponential (SE), and bi-exponential (BE) components, respectively. HSGDNet accelerates whole-knee T1ρ fitting over NLS by approximately 67.4 × for ME, 53.9 × for SE, and 42.3 × for BE. Finally, to evaluate robustness under pathological and protocol variations, we validate HSGDNet on an early osteoarthritis (EOA) dataset acquired with distinct spin-lock times (TSLs) values. Overall, HSGDNet establishes itself as an efficient method for rapid, precise, and robust multi-component T1ρ mapping in the knee joint.

Bone is invisible with conventional MRI sequences. It is highly desirable to develop novel MRI sequences to image bone, providing a radiation-free modality for skeletal imaging. We compared the morphological and quantitative strength of three-dimensional (3D) ultrashort echo time (UTE), zero echo time (ZTE), and short TR adiabatic inversion recovery UTE (STAIR-UTE) MRI techniques for bone imaging in various skeletal anatomical regions, including the forearm, wrist, lower leg, upper leg, and skull. Five healthy volunteers (four male and one female) were subject to four MRI sequences, including 3D UTE with 2° and 7° flip angles (FAs), 3D ZTE with 2° FA, and 3D STAIR-UTE with 14° FA. Regions of interest (ROIs) were drawn in cortical bone, marrow cavity, and muscle to measure their signal intensities. An artifact-free ROI was also placed in the image background to measure the standard deviation (SD) of noise. The signal-to-noise ratio of bone (SNR_Bone) and contrast-to-noise ratios (CNRs) between bone and marrow (CNR_Bone-Marrow) and bone and muscle (CNR_Bone-Muscle) were measured in different anatomical regions. These SNR and CNRs were divided by the square root of acquisition time. In addition, bone volume renderings were generated from 3D STAIR-UTE images. The averages and SDs of SNR_Bone, CNR_Bone-Marrow, and CNR_Bone-Muscle were calculated for different anatomical regions. UTE with 7° FA has the highest positive SNR and negative CNR. UTE and ZTE sequences with the same FAs of 2° have similar SNR and CNR values. The STAIR-UTE sequence with 14° FA has the lowest SNR but is the only sequence providing positive CNR for bone at all investigated body regions, which can be used for direct bone volume rendering. The STAIR-UTE technique provides high contrast volumetric imaging of skeletal anatomies, which enables us to generate direct bone surface-rendered images in clinically acceptable scan time.

With the increasing adoption of ultrahigh-field MRI scanners, there is a growing interest in migrating MRS studies from 3 to 7 T. Prior field comparisons of MRS in healthy volunteers demonstrated better reliability of quantification at 7 T, particularly for weakly represented metabolites. Neurochemical quantification has not been compared at 3 T versus 7 T in clinical cohorts and under controlled physiological conditions. In this exploratory study, we analyzed MRS data from the hypothalamus and prefrontal cortex volumes of interest (VOIs) that were collected from the same individuals with Type 1 diabetes at 3 and 7 T as a part of a larger study investigating cerebral responses to glycemic changes. Seventeen individuals underwent MRS during euglycemic clamps at both 3 T and 7 T, allowing us to compare metabolite concentrations obtained at the two fields with a consensus-recommended short-echo semi-LASER protocol under the same physiological conditions. Our aim was to examine whether there are systematic biases in neurochemical concentrations measured at 3 T versus 7 T and to assess whether creatine (tCr) ratios would reduce or eliminate such biases. High-quality spectra were obtained from both VOIs and fields, with 8-15 reliably quantified (mean Cramér-Rao lower bounds ≤ 20%) metabolites from LCModel. Whereas neurochemical profiles were highly similar between 3 and 7 T, several metabolites exhibited systematic differences with water-referenced quantifications, such as glucose + taurine (Glc + Tau) and phosphoethanolamine. Quantifications with tCr as reference did not alleviate the biases and, in fact, resulted in a larger number of significant differences due to systematic biases in the tCr concentration. Pearson correlation analysis showed significant associations for several metabolites between 3 and 7 T, suggesting that interindividual differences in neurochemical levels are detectable. Associations were stronger when using tCr ratios. Importantly, hypothalamic Glc + Tau showed a strong correlation between fields at these tightly regulated euglycemic conditions, opening the possibility to detect individualized glucose concentrations in this brain region that participates in the regulation of blood glucose levels.

Investigating glucose metabolism in the brain using [6,6-²H₂]glucose (²H₂-Glc) and deuterium-based NMR spectroscopy has shown promise for noninvasive monitoring of the fate of this labeled compound. This approach has already been applied in vivo in small animals and human subjects. A model of perfused rat brain slices recently showed promise for the investigation of the metabolic consequences of acute ischemic stroke, which is a significant cause of death and morbidity worldwide. The current study aimed to implement the deuterium-based glucose metabolism monitoring approach to study the metabolic consequences of ischemia and reperfusion in the rat brain ex vivo. In agreement with previous studies, we found that deuterated lactate (²H₂-Lac) was immediately formed in the brain upon administration of ²H₂-Glc to the perfusion medium. This metabolite remained the predominant metabolic fate observed in the ²H-NMR spectra. Upon perfusion arrest, ²H₂-Lac quickly built up to the same amount of ²H₂-Glc eliminated from the medium engulfing the slices, reaching fivefold to sixfold its baseline level (n = 6, three animals, and two ischemic conditions in each). Upon reperfusion, ²H₂-Lac decreased to its level before the ischemic condition, and ²H₂-Glc returned to its baseline. ²H₂-Lac washout to the medium amounted to 2.2% of the ²H₂-Lac signal associated with the slices after about 5 h of perfusion with ²H₂-Glc, suggesting that the ²H₂-Lac signal observed during the experiments was predominantly intracellular. These results demonstrate the utility of ²H₂-Glc and ²H-NMR in monitoring the consequences of ischemia and reperfusion in the perfused rat brain slices model.

Mild traumatic brain injury (mTBI) caused by sports-related incidents in children and youth often leads to prolonged cognitive impairments but remains difficult to diagnose. In order to identify clinically relevant imaging and behavioral biomarkers associated concussion, a closed-head mTBI was induced in adolescent pigs. Twelve (n = 4 male and n = 8 female), 16-week old Yucatan pigs were tested; n = 6 received mTBI and n = 6 received a sham procedure. T1-weighted imaging was used to assess volumetric alterations in different regions of the brain and diffusion tensor imaging (DTI) to examine microstructural damage in white matter. The pigs were imaged at 1- and 3-month post-injury. Neuropsychological screening for executive function and anxiety were performed before and in the months after the injury. The volumetric analysis showed significant longitudinal changes in pigs with mTBI compared with sham, which may be attributed to swelling and neuroinflammation. Fractional anisotropy (FA) values derived from DTI images demonstrated a 21% increase in corpus callosum from 1 to 3 months in mTBI pigs, which is significantly higher than in sham pigs (4.8%). Additionally, comparisons of the left and right internal capsules revealed a decrease in FA in the right internal capsule for mTBI pigs, which may indicate demyelination. The neuroimaging results suggest that the injury had disrupted the maturation of white and gray matter in the developing brain. Behavioral testing showed that compare to sham pigs, mTBI pigs exhibited 23% increased activity in open field tests, 35% incraesed escape attempts, along with a 65% decrease in interaction with the novel object, suggesting possible memory impairments and cognitive deficits. The correlation analysis showed an associations between volumetric features and behavioral metrics. Furthermore, a machine learning model, which integrated FA, volumetric features and behavioral test metrics, achieved 67% accuracy, indicating its potential to differentiate the two groups. Thus, the imaging biomarkers were indicative of long-term behavioral impairments and could be crucial to the clinical management of concussion in youth.

Recent evidence has demonstrated that several white matter (WM) microstructural features, such as axon diameter, fibre configurations and fibre orientation in respect to the magnetic field influence T1 relaxation. The effects from microstructural features on T1 are small in size, thus, visualising the effects of WM microstructure remains challenging in standard T1 weighted MRI in vivo. Here, we have studied an algebraic approach involving subtraction, addition and division of closely spaced inversion time images in WM imaging, the so-called dSIR approach. Images collected with short TI (300 ms at 3T and 600 ms at 7T) and long TI (600 ms at 3T and 1000 ms at 7T) with MP2RAGE MRI were combined using the dSIR processing. dSIR signal intensities were compared with absolute T1 images. We found that dSIR was linearly related with T1 relaxation time over approximately 200 ms both at 3T and 7T. The slope of the dSIR versus T1 plot was 1.6 times greater at 7T than at 3T indicative of higher dSIR contrast at 7T. dSIR contrast revealed WM tracts that are oriented with high angle (fibre-to-field angle > 75°), in addition, dSIR signal showed angular patterns that closely resembled those of T1 at both fields. The dSIR contrast due to intratissue T1 difference of order of ~50 ms generated by microstructural features, including axon fibre orientation as well as by the presence of large and giant axons in somato-motor subsection of corpus callosum were visualised. It is concluded that dSIR signal mimics T1 and that the dSIR contrast is higher at 7T than at 3T; thus, the approach will help to visualise the effects of microstructure on T1 to evaluate WM integrity.

Magnetic resonance fingerprinting (MRF), as an emerging versatile and noninvasive imaging technique, provides simultaneous quantification of multiple quantitative MRI parameters, which have been used to detect changes in cartilage composition and structure in osteoarthritis. Deep learning (DL)-based methods for quantification mapping in MRF overcome the memory constraints and offer faster processing compared to the conventional dictionary matching (DM) method. However, limited attention has been given to the fine-tuning of neural networks (NNs) in DL and fair comparison with DM. In this study, we investigate the impact of training parameter choices on NN performance and compare the fine-tuned NN with DM for multiparametric mapping in MRF. Our approach includes optimizing NN hyperparameters, analyzing the singular value decomposition (SVD) components of MRF data, and optimization of the DM method. We conducted experiments on synthetic data, the NIST/ISMRM MRI system phantom with ground truth, and in vivo knee data from 14 healthy volunteers. The results demonstrate the critical importance of selecting appropriate training parameters, as these significantly affect NN performance. The findings also show that NNs improve the accuracy and robustness of T₁, T₂, and T_1ρ mappings compared to DM in synthetic datasets. For in vivo knee data, the NN achieved comparable results for T₁, with slightly lower T₂ and slightly higher T_1ρ measurements compared to DM. In conclusion, the fine-tuned NN can be used to increase accuracy and robustness for multiparametric quantitative mapping from MRF of the knee joint.

Magnetic resonance spectroscopy (MRS) enables noninvasive quantification of metabolites, but its utility in vivo can be limited by low signal-to-noise ratios (SNRs) and long acquisition times. The use of ultrahigh-field (UHF) strengths (> 3 T) combined with multichannel phased receive arrays can improve spectral SNR. A crucial step in the use of multichannel arrays is the combination of spectra acquired from individual coil channels. We previously developed a coil combination method at 3 T, optimized truncation to integrate multichannel MRS data using rank-R singular value decomposition (OpTIMUS), which uses noise-whitened windowed spectra and iterative rank-R singular value decomposition (SVD) to combine multichannel MRS data. Here, we evaluated OpTIMUS for combination of MR spectra acquired using a multichannel phased array at 7 T and compared spectral SNR and metabolite quantification with spectra combined using whitened SVD (WSVD), signal/noise squared (S/N²), and the Brown method. Data were acquired from 14 healthy volunteers, including five with data acquired at both 3 and 7 T, and from nine people living with HIV. Spectra combined using OpTIMUS resulted in a higher SNR compared to the three other methods, consistent with our prior results at 3 T. With half the number of averages (N = 32), spectra combined with OpTIMUS had higher SNR compared to spectra using the Brown method with 64 averages. Additionally, spectra combined using OpTIMUS at 7 T were compared to spectra acquired at 3 T with the same number of averages (N = 64) or matched acquisition times (N = 110 averages), and spectral fitting was consistently improved at 7 T even when comparable SNR was obtained at 3 T. The ability to increase SNR and maintain spectral quality by optimizing spectral coil combination has the potential to reduce scan time, a key challenge for routine clinical use of MRS.