Magnetic Resonance Materials in Physics, Biology and Medicine最新文献

Background: Free-running whole-heart MRI using balanced steady-state free precession (bSSFP) sequences offer high SNR and myocardial tissue contrast. However, an inadequate fat signal suppression may introduce artifacts and is particularly challenging with non-Cartesian readouts. The aim of this study was to evaluate different fat-signal suppression methods for whole-heart free-running MRI at 1.5 T using numerical simulations, phantom, and cardiac MRI experiments without the use of contrast agents.

Methods: Binomial off-resonant rectangular (BORR), lipid insensitive binomial off-resonant RF excitation (LIBRE), and lipid insensitive binomial off-resonant (LIBOR) pulses were implemented within a 3D radial bSSFP sequence. Their pulse parameters were optimized for fat signal suppression at 1.5 T using simulations and phantom experiments. Optimized protocols, along with a free-running fast interrupted steady-state (FISS) and non-fat suppressed bSSFP sequence, were used to acquire phantom and cardiac data in five volunteers. SAR values were recorded. The SNR and CNR_Water-Fat were measured in phantom data, while SNR and CNR_{Blood-Myocardium} were quantified in volunteers using reconstruction without motion correction. Motion-resolved reconstructions were used for qualitative assessments. Statistical differences were analyzed using one-way ANOVA.

Results: LIBOR had the highest CNR_Water-Fat (276.8 ± 2.5) in phantoms, followed by LIBRE (268.1 ± 2.6), BORR (249.9 ± 2.2), and FISS (212.7 ± 2.7), though these differences were not statistically significant (p > 0.05). In volunteers, BORR had the highest SNR in the ventricular blood pool (17.0 ± 1.5), and LIBRE had the highest CNR_Blood-Fat (29.4 ± 9.3). FISS had the highest CNR_{Blood-Myocardium} (29.0 ± 8.9), but the differences were not significant (p > 0.05). Motion-resolved cardiac imaging showed comparable quality across all fat-suppressed sequences, with no significant streaking artifacts observed. Free-running bSSFP with LIBOR required the lowest SAR, up to a sixfold decrease compared with FISS.

Conclusion: The tested sequences performed similarly in SNR and CNR but LIBOR offered the lowest SAR, making it a promising candidate for applications where RF energy deposition is a concern.

Objective: Reliable image quality assessment is crucial for evaluating new motion correction methods for magnetic resonance imaging. We compare the performance of common reference-based and reference-free image quality metrics on unique datasets with real motion artifacts, and analyze the metrics' robustness to typical pre-processing techniques.

Materials and methods: We compared five reference-based and five reference-free metrics on brain data acquired with and without intentional motion (2D and 3D sequences). The metrics were recalculated seven times with varying pre-processing steps. Spearman correlation coefficients were computed to assess the relationship between image quality metrics and radiological evaluation.

Results: All reference-based metrics showed strong correlation with observer assessments. Among reference-free metrics, Average Edge Strength offers the most promising results, as it consistently displayed stronger correlations across all sequences compared to the other reference-free metrics. The strongest correlation was achieved with percentile normalization and restricting the metric values to the skull-stripped brain region. In contrast, correlations were weaker when not applying any brain mask and using min-max or no normalization.

Discussion: Reference-based metrics reliably correlate with radiological evaluation across different sequences and datasets. Pre-processing significantly influences correlation values. Future research should focus on refining pre-processing techniques and exploring approaches for automated image quality evaluation.

Purpose: Quantitative MRI markers, such as myelin water fraction (MWF) and geometric mean $T_2$ (IET2) (the intra-/extra-cellular water compartment), can be biomarkers for various brain disorders. However, these markers require acquiring multi-echo spin-echo images which requires long scan times. Undersampled 3D-GRAdient Echo and Spin Echo (3D-GRASE) scans with parallel imaging have been used for faster scans. Still, further acceleration is desirable. Reconstruction techniques that utilize redundancy along the echoes could be employed to achieve artifact-free maps at higher acceleration. This work examines the possibility of using one such technique, subspace constrained reconstruction (SCR), for further accelerating the 3D-GRASE scan.

Methods: We propose two techniques to undersample the 3D-GRASE acquisition and exploit the redundancy across echoes. We retrospectively undersample fully sampled data from phantom and in-vivo acquisition to test these techniques. We compared our results for mapping MWF and IET2 to a reference multi-spin-echo technique. Additionally, we compare the proposed, state-of-the-art, and reference techniques with prospectively undersampled in-vivo acquisitions.

Results: The RMSD of the MWF in retrospectively undersampled data was worse for the proposed techniques than the state-of-the-art. However, for IET2, RMSD was similar or slightly improved. In prospectively undersampled scans, undersampling artifacts deteriorated MWF maps, but not IET2 maps, which were within 10 ms of the reference map.

Conclusion: Our findings suggest that exploiting redundancy across echoes does not result in additional acceleration beyond the current state-of-the-art for MWF mapping, while it is possible to accelerate beyond state-of-the-art for IET2 mapping.

Objective: To evaluate the effects of excluding fatty tissue in QSM of human knee cartilage.

Materials and methods: Gradient echo images from 18 knee-healthy volunteers were acquired, from which chemical shift corrected field perturbation maps were calculated. Based on these, QSM maps were reconstructed using morphology enabled dipole inversion and one of three masking alternatives: (1) excluding no tissue, (2) excluding bone marrow, and (3) excluding all fatty tissues. The slope of a linear regression [ppm/%] between susceptibility values and the relative distance from the bone surfaces was used as a measurement of contrast between cartilage layers. The average differences in slopes between methods are reported with 95% confidence intervals.

Results: The expected susceptibility differences between cartilage layers from literature were observed for all tested reconstruction techniques. However, smaller slopes (average difference (confidence interval)) were detected when either all fatty tissue (- 0.090 (- 0.121, - 0.059) ppm/%) or bone marrow (- 0.088 (- 0.121, - 0.055) ppm/%) was excluded from reconstruction.

Discussion: All tested methods result in adequate image quality in QSM of knee cartilage. However, exclusion of fatty tissue decreased the susceptibility contrast between cartilage layers. Assuming that phase contributions from chemical shift are addressed, inclusion of fatty tissue may be preferable.

Background: Proton density fat fraction (PDFF)- the ratio of unconfounded fat signal to the sum of the unconfounded fat and water signals, is a valuable quantitative imaging biomarker of metabolic associated steatotic liver disease (MASLD) widely applied in clinical practice and clinical trials. PDFF of the liver is commonly measured using 3 T MRI systems. However, low-field systems are increasingly favored due to lower cost, improved safety profile, minimized artifacts around metallic implants, and enhanced patient comfort.

Objective: In this pilot study, we used knowledge of standardized and widely used 3 T liver PDFF protocols, and adapted parameters to be appropriate for the 0.55 T MRI. We evaluate a liver fat quantification protocol at 0.55 T compared to a standard clinical 3 T protocol to measure liver fat in patients with MASLD.

Material and methods: Eight adult patients (average age 53.6 ± 13.6 years, 5 females) with ≥ 5% PDFF on 3 T MRI underwent a 0.55 T MRI PDFF protocol within 90 days. To keep the acquisition time to be within a reasonable breath hold duration and with reasonable signal-to-noise ratio (SNR), four echoes were acquired at a lower resolution and fewer number of slices at 0.55 T compared to 3 T which uses a 6-echo multi-echo Dixon volumetric interpolated breath hold examination (VIBE) protocol. PDFF quantification accuracy of the 0.55 T approach was evaluated using a commercial PDFF phantom and in vivo.

Results: In the phantom, there was excellent match (R² > 0.999) between PDFF estimated by 0.55 T MRI and ground truth. Mean in vivo 3 T MRI-PDFF was 16.5%, compared to 16.3% 0.55 T MRI-PDFF (correlation coefficient r = 0.99). The Bland-Altman analysis showed good agreement of in vivo PDFF measurements across 0.55 T and 3 T estimating a bias or mean difference of - 0.25% and the limits of agreements (LoA) of - 3.98% and 3.48%.

Discussion: Our data demonstrate that 0.55 T MRI is feasible and comparable to 3 T MRI in quantifying liver PDFF among patients with MASLD.

Objectives: Developing a ¹⁹F imaging method to acquire images of the molecular inflammation tracer perfluorooctyl bromide (PFOB) without chemical shift artifacts.

Materials and methods: PFOB is a molecular tracer that can be used to track the response of myeloid cells. However, imaging of PFOB with ¹⁹F-MRI is challenging due to its complex spectrum which leads to unwanted chemical shift artifacts. Spectral HE allows for separate reconstructions of each peak of the PFOB spectrum, which was combined into a single image after resonance shift correction. In this work, a Hadamard-encoded (HE) radial 3D UTE sequence was tested in phantoms and in vivo in a pig, measuring the ¹⁹F signal in the spleen at different times after injection.

Results: Chemical shift artifacts were effectively suppressed with HE, and an SNR > 100 was observed for the ¹⁹F signal in the spleen 2 days after injection. The signal decreased over time, and 7 days after injection it was reduced by 30%.

Discussion: Chemical shift artifact correction using HE allowed for in vivo ¹⁹F PFOB imaging of labeled monocytes with a high SNR. Compared to spectrally selective excitation, HE increased the PFOB ¹⁹F-MRI signal by 10%, and the simple HE-algorithm could be directly integrated into the image reconstruction of the MRI system.

Objective: To reduce errors from J-modulations and spectral overlap in dMRS of brain metabolites, this study combines the use of diffusion-weighted gradients with selective refocusing and spectral editing.

Materials and methods: Bipolar gradients were combined with spectral refocusing and editing in a dMEGA-PRESS sequence. Experimental parameters were optimised for spectral editing of GABA, with co-editing of Glutamate and Glutamine. The method was tested in metabolite phantom solutions, followed by pre-clinical experiments on rats.

Results: The dMEGA-PRESS sequence enabled reliable spectral editing and quantification of GABA. Selective refocusing and editing resulted in reduced uncertainty in the diffusion data for GABA and Glutamate in the metabolite phantoms, and also for the combined Glutamate/Glutamine diffusion data obtained in vivo. Reliable diffusion data for GABA was not possible to obtain from the in vivo spectra.

Discussion: For metabolites with significant J-modulations but without spectral overlap, selective refocusing improved the quality of diffusion data. For metabolites with spectral overlap where editing is necessary, spectral subtraction makes it more challenging to improve the quality of diffusion-weighted data.

Conclusion: The dMEGA-PRESS sequence reduces the uncertainty in obtained diffusion data for brain metabolites that are significantly influenced by J-modulations.

Objective: Deep complex convolutional networks (DCCNs) utilize complex-valued convolutions and can process complex-valued MRI signals directly without splitting them into two real-valued magnitude and phase components. The performance of DCCN and real-valued U-Net is thoroughly investigated in the physics-informed subject-specific ad-hoc reconstruction method for fat-water separation and is compared against a widely used reference approach.

Materials and methods: A comprehensive test dataset (n = 33) was used for performance analysis. The 2012 ISMRM fat-water separation workshop dataset containing 28 batches of multi-echo MRIs with 3-15 echoes from the abdomen, thigh, knee, and phantoms, acquired with 1.5 T and 3 T scanners were used. Additionally, five MAFLD patients multi-echo MRIs acquired from our clinical radiology department were also used.

Results: The quantitative results demonstrated that DCCN produced fat-water maps with better normalized RMS error and structural similarity index with the reference approach, compared to real-valued U-Nets in the ad-hoc reconstruction method for fat-water separation. The DCCN achieved an overall average SSIM of 0.847 ± 0.069 and 0.861 ± 0.078 in generating fat and water maps, respectively, in contrast the U-Net achieved only 0.653 ± 0.166 and 0.729 ± 0.134. The average liver PDFF from DCCN achieved a correlation coefficient R of 0.847 with the reference approach.

Objective: Texture analysis in quantitative IVIM-DKI parameters was investigated for characterizing malignant and benign lymph nodes and distinguishing lymphoma subtypes, specifically Hodgkin lymphoma (HL) and non-Hodgkin lymphoma (NHL).

Methods: A prospective cohort of twenty-one patients (n = 21) diagnosed with biopsy-proven lymphoma (HL: 13 and NHL: 8) were analyzed. All patients underwent conventional MRI including DWI with 9 b-values (0-2000s/mm²) at 1.5 T and whole-body FDG-PET/CT. IVIM-DKI parameters were estimated using IVIM-DKI model with total-variation (TV) spatial-regularization method (IDTV). Apparent diffusion coefficient (ADC) and standard uptake value (SUV) maps were also calculated. Total 31 of 3D texture features using global and second-order textures were extracted from imaging parameters in the volume-of-interest of malignant and benign lymph nodes. Machine learning linear classifier model was developed for distinguishing between malignant from benign lymph nodes and HL from NHL using textural features and area under receiver operating curve (AUROC) that were evaluated to assess diagnostic accuracy.

Results: Texture parameters of neighborhood gray-tone difference matrix (NGTDM) in all IVIM-DKI parameters along with ADC demonstrated excellent diagnostic accuracy showing the highest AUROC of 0.99 (individual highest AUROC by ADC: 0.99; AUROC by all: 0.95-0.99) for distinguishing between malignant and benign lymph nodes. While gray-level co-occurrence matrix (GLCM) and gray-level run-length matrix (GLRLM) features in ADC, diffusion coefficient (D), perfusion coefficient (D*), and perfusion fraction (f) displayed the best AUROC of 0.98 (individual highest AUROC by D: 0.96; AUROC by all: 0.85-0.96) for distinguishing HL from NHL.

Conclusion: Texture analysis of IVIM-DKI parameters showed promising diagnostic performance in characterizing HL and NHL. Quantitative IVIM-DKI analysis with TV may have a wide range of applicability for the clinical evaluation of lymphomas.