Massively multidimensional diffusion-relaxation correlation MRI provides detailed information on tissue microstructure by analyzing water populations at a sub-voxel level. This method correlates frequency-dependent tensor-valued diffusion MRI with longitudinal and transverse relaxation rates, generating non-parametric D(ω)-R 1-R 2-distributions. Traditionally, D(ω)-R 1-R 2-distributions are separated using manual binning of the diffusivity and anisotropy space to differentiate white matter (WM), gray matter (GM), and free water (FW) in brain tissue. However, while effective, this approach oversimplifies complex tissue fractions and does not fully utilize all available diffusion-relaxation parameters. In this study, we implemented an unsupervised clustering approach to automatically classify WM, GM, and FW and explore additional water populations using all components in the D(ω)-R 1-R 2-distributions on ex vivo and in vivo rat brain, and in vivo human brain. Results showed that a basic separation of WM, GM, and FW is possible using unsupervised clustering even under different multidimensional diffusion-relaxation protocols of rat brain and human brain. Additionally, when there is high frequency-dependent diffusion range, it is possible to obtain a cluster characterized by restriction localized in specific high cell density regions such as the dentate gyrus and cerebellum of rat brain. These findings were compared with rat histological sections of myelin and Nissl stainings. We demonstrated that unsupervised clustering of diffusion-relaxation MRI data can reveal tissue complexity beyond traditional WM, GM, and FW segmentation in rat and human brain without parameter assumptions. The unsupervised cluster approach could be used in other body parts (e.g., prostate and breast cancer) without requiring pre-defined bin limits. Furthermore, the characterization of the clusters by diffusivities, anisotropy, and relaxation rates can provide a better understanding of the subtle changes in different cellular fractions in tissue-specific pathologies.

Abstract Alzheimer’s disease (AD) pathology involves amyloid deposition, reactive gliosis, and localized tissue alterations that coexist within the same brain regions, creating heterogeneous microstructural environments within individual imaging voxels. Conventional diffusion MRI averages these environments into aggregate measures, potentially obscuring their distinct contributions. Frequency-dependent multidimensional MRI (ωMD-MRI) resolves distributions of water components with different diffusion length scales, anisotropies, and relaxation properties, providing sensitivity to microstructural restriction, heterogeneity, and shape-size correlations within a voxel. Whether these measurements reveal microstructural complexity associated with AD pathology remains unclear. Here, we performed ωMD-MRI on ex vivo brain specimens from approximately 8-month-old 5xFAD and wild-type mice and interpreted the imaging findings alongside complementary histology. ωMD-MRI revealed widespread but spatially nonuniform differences between 5xFAD and wild-type brains. Measurements sensitive to microstructural restriction, heterogeneity, and shape-size correlations consistently indicated greater microstructural heterogeneity in 5xFAD brains, with the most prominent differences in the hippocampal formation and major cerebral white matter tracts. Complementary qualitative histology demonstrated extensive amyloid deposition and glial activation in affected regions, while overall cytoarchitecture and myelin organization remained largely preserved. Thus, the ωMD-MRI abnormalities occurred in tissue characterized by multiple coexisting pathological and relatively preserved microstructural environments rather than widespread structural degeneration. These findings demonstrate that ωMD-MRI can reveal the spatial and microstructural heterogeneity associated with amyloid pathology and provide a more comprehensive characterization of AD-related tissue alterations.

Frequency-dependent diffusion-relaxation distribution MRI provides information beyond the traditional voxel-averaged metric that may better characterize the microstructural features of biological tissue. Frequency-dependent multidimensional ( ω $$ \omega $$ MD) MRI reproducibility has been established in clinical settings, but has yet to be thoroughly evaluated under preclinical conditions, where superior hardware and modulated gradient waveforms enhance its performance. In this study, we investigate the reproducibility of ω $$ \omega $$ MD-MRI using a micro-imaging system to investigate ex vivo mouse brains. Notably, the estimated signal fractions of intra-voxel spectral components in the ω $$ \omega $$ MD-MRI distribution, corresponding to white and gray matter, along with the frequency-dependent parameters, demonstrated high reproducibility. We identified bias between scan and rescan in some of the metrics, which we attribute to the time gap between repeated scans pointing to a long-time progressive fixation effect. We compare our results with in vivo results from clinical scanners and show the reproducibility of diffusion frequency-dependent metrics to benefit from the improved gradient hardware on our preclinical setup. Our results inform future micro-imaging ex vivo studies of the reproducibility of ω $$ \omega $$ MD-MRI metrics and their dependence on fixation time.

Abstract Quantitative microstructural magnetic resonance imaging (MRI) can noninvasively characterize tissue configuration at micrometer scales, but clinical uptake is limited by validation and optimization in human-relevant scenarios. Organoids are powerful human-relevant tissue models, yet translation is hampered by lack of non-destructive, longitudinal microstructural assessment. Bridging these gaps, microstructural MRI of living organoids can accelerate MRI biomarker and organoid development and validation. Here, we address key obstacles to enable organoid microstructural MRI. First, we use a unique 28.2 T MRI system to achieve spatial resolution with adequate signal-to-noise ratio and feasible scan times. Second, we implement flexible acquisitions with fast readouts to expand multivariate experimental capacity. Third, we develop a workflow combining 3D MRI and 3D lightsheet microscopy for cross-modality anatomical comparison beyond 2D. Using this platform, we demonstrate microstructural MRI of cortical organoids with resolutions down to (20 µm ) 3 , revealing anisotropy, heterogeneity, maturation-dependent differences, and temporal changes in cortical organoids. Correlative lightsheet microscopy confirms correspondence to axonal and nuclear architecture. This platform enables live-organoid MRI as a complementary tool to human- and animal imaging for robust microstructural assessment.

Non-invasive MRI is widely used to assess and monitor ischemic stroke, yet conventional approaches often lack sensitivity to subtle microstructural changes and struggle to evaluate tissue viability across lesion, penumbra, and distal regions. In this study, frequency dependent diffusion tensor distribution imaging (ωDTD) was combined with clustering of diffusion tensor distributions D(ω) and multivariate regression modeling to characterize ischemic tissue alterations in a whole brain section. Ex vivo ωDTD and histology were performed in rats subjected to middle cerebral artery occlusion (MCAO) or sham surgery (P = 17) 24 hours after reperfusion. Lesions showed cell loss and an increased presence of smaller, likely glial, cells. A random forest (RF) model was used to explain and predict histological parameters from diffusion tensor imaging (DTI), manually bin resolved ωDTD features, and cluster resolved ωDTD parameters. Model performance was evaluated using leave one animal out cross validation (LOO CV). ωDTD features better represented cell number than DTI metrics (ωDTD R2 = 0.73 vs. DTI R2 = 0.49), with similar advantages for nuclear area and circularity (ωDTD R2 = 0.64 and 0.61 vs. DTI R2 = 0.40 and 0.35). The RF model further proved beneficial in capturing complex, nonlinear relationships between MRI parameters and tissue characteristics. Overall, these results indicate that ωDTD provides richer microstructural information than standard DTI, and that combining ωDTD with advanced machine learning methods enhances interpretation of ischemic tissue damage.

Massively multidimensional diffusion-relaxation correlation MRI (MMD-MRI) provides information beyond the traditional voxel-averaged metric that may better characterize the microstructural characteristics of biological tissues. MMD-MRI reproducibility has been established in clinical settings, but has yet to be thoroughly evaluated under preclinical conditions, where superior hardware and modulated gradient waveforms enhance its performance. In this study, we investigate the reproducibility of MMD-MRI on a micro-imaging system using ex vivo mouse brains. Notably, the estimated signal fractions of intra-voxel spectral components in the MD-MRI distribution, corresponding to white and gray matter, along with the frequency-dependent parameters, demonstrated high reproducibility. We identified bias between scan and rescan in some of the metrics, which we attribute to the time gap between repeated scans pointing to a long-time progressive fixation effect. We compare our results with in vivo results from clinical scanners and show the reproducibility of diffusion frequency-dependent metrics to benefit from the improved gradient hardware on our preclinical setup. Our results inform future micro-imaging ex vivo MMD-MRI studies of the reproducibility of MMD-MRI metrics and their dependence on fixation time.

Time- or frequency-dependent ("restricted") diffusion potentially provides useful information about cellular-scale structures in the brain but is challenging to interpret because of intravoxel tissue heterogeneity. Multidimensional diffusion-relaxation correlation MRI with tensor-valued diffusion encoding enables characterization of intravoxel heterogeneity in terms of nonparametric distributions of diffusion tensors and nuclear relaxation rates, and was recently augmented with explicit consideration of frequency-dependence to resolve the effects of restricted diffusion for distinct populations of tissue water. The simplest acquisition protocols for tensor-valued encoding unintentionally cover a frequency range of a factor 2-3, which can be extended in a more controlled way with oscillating gradient waveforms. While microimaging equipment with high-amplitude magnetic field gradients allows exploration of frequencies from tens to hundreds of Hz, clinical scanners with more moderate gradient capabilities are limited to narrower ranges that may be insufficient to observe restricted diffusion for brain tissues. We here investigate the effects of including or omitting frequency-dependence in the data inversion from isotropic and anisotropic liquids, excised tumor tissue, ex vivo mouse brain, and in vivo human brain. For microimaging measurements covering a wide frequency range, from 35 to 320 Hz at b-values over 4·109 sm-2, the inclusion of frequency-dependence drastically reduces fit residuals and avoids bias in the diffusion metrics for tumor and brain voxels with micrometer-scale structures. Conversely, for the case of in vivo human brain investigated in the narrow frequency range from 5 to 11 Hz at b = 3·109 sm-2, analyses with and without inclusion of frequency-dependence yield similar fit residuals and diffusion metrics for all voxels. These results indicate that frequency-dependent inversion may be generally applied to diffusion-relaxation correlation MRI data with and without observable effects of restricted diffusion.

The knowledge of the cardiac microstructure and the 3D myofiber architecture grow years after years with the multiplication and the upgrade of imaging technologies. However, the course of events of pathophysiological processes like cardiac remodeling, and the link with clinical phenotypes are not yet clearly understood. Some concerns have been raised regarding the interpretation of the late gadolinium enhancement (LGE) at the right ventricle attachment or insertion point (RVIP) however 3D microstructure organization of the RVIP has not been extensively described in the literature.

Magnetic resonance imaging (MRI) is the method of choice for noninvasive studies of micrometer-scale structures in biological tissues via their effects on the time- and frequency-dependent (restricted) and anisotropic self-diffusion of water. While new designs of time-dependent magnetic field gradient waveforms have enabled disambiguation between different aspects of translational motion that are convolved in traditional MRI methods relying on single pairs of field gradient pulses, data analysis for complex heterogeneous materials remains a challenge. Here, we propose and demonstrate nonparametric distributions of tensor-valued Lorentzian diffusion spectra, or “D(ω) distributions,” as a general representation with sufficient flexibility to describe the MRI signal response from a wide range of model systems and biological tissues investigated with modulated gradient waveforms separating and correlating the effects of restricted and anisotropic diffusion.

Massively multidimensional diffusion magnetic resonance imaging combines tensor-valued encoding, oscillating gradients, and diffusion-relaxation correlation to provide multicomponent subvoxel parameters depicting some tissue microstructural features. This method was successfully implemented ex vivo in microimaging systems and clinical conditions with tensor-valued gradient waveform of variable duration giving access to a narrow diffusion frequency ( ω ) range. We demonstrate here its preclinical in vivo implementation with a protocol of 389 contrast images probing a wide diffusion frequency range of 18 to 92 Hz at b -values up to 2.1 ms/μm 2 enabled by the use of modulated gradient waveforms and combined with multislice high-resolution and low-distortion echo planar imaging acquisition with segmented and full reversed phase-encode acquisition. This framework allows the identification of diffusion ω -dependence in the rat cerebellum and olfactory bulb gray matter (GM), and the parameter distributions are shown to resolve two water pools in the cerebellum GM with different diffusion coefficients, shapes, ω -dependence, relaxation rates, and spatial repartition whose attribution to specific microstructure could modify the current understanding of the origin of restriction in GM.

While the microstructure of the left ventricle (LV) has been largely described, only a few studies investigated the right ventricular insertion point (RVIP). It was accepted that the aggregate cardiomyocytes organization was much more complex due to the intersection of the ventricular cavities but a precise structural characterization in the human heart was lacking even if clinical phenotypes related to right ventricular wall stress or arrhythmia were observed in this region. MRI-derived anatomical imaging (150 µm3) and diffusion tensor imaging (600 µm3) were performed in large mammalian whole hearts (human: N = 5, sheep: N = 5). Fractional anisotropy, aggregate cardiomyocytes orientations and tractography were compared within both species. Aggregate cardiomyocytes orientation on one ex-vivo sheep whole heart was then computed using structure tensor imaging (STI) from 21 µm isotropic acquisition acquired with micro computed tomography (MicroCT) imaging. Macroscopic and histological examination were performed. Lastly, experimental cardiomyocytes orientation distribution was then compared to the usual rule-based model using electrophysiological (EP) modeling. Electrical activity was modeled with the monodomain formulation. The RVIP at the level of the inferior ventricular septum presented a unique arrangement of aggregate cardiomyocytes. An abrupt, mid-myocardial change in cardiomyocytes orientation was observed, delimiting a triangle-shaped region, present in both sheep and human hearts. FA’s histogram distribution (mean ± std: 0.29 ± 0.06) of the identified region as well as the main dimension (22.2 mm ± 5.6 mm) was found homogeneous across samples and species. Averaged volume is 0.34 cm3 ± 0.15 cm3. Both local activation time (LAT) and morphology of pseudo-ECGs were strongly impacted with delayed LAT and change in peak-to-peak amplitude in the simulated wedge model. The study was the first to describe the 3D cardiomyocytes architecture of the basal inferoseptal left ventricle region in human hearts and identify the presence of a well-organized aggregate cardiomyocytes arrangement and cardiac structural discontinuities. The results might offer a better appreciation of clinical phenotypes like RVIP-late gadolinium enhancement or uncommon idiopathic ventricular arrhythmias (VA) originating from this region.

A novel thermometry acquisition and a fast deep learning based image reconstruction were combined for cardiac interventional thermometry at high spatial (0.86×0.86mm 2 ) and temporal (0.97s) resolutions, robust to motion and susceptibility artefact and independent of external ECG-gating. The method was tested in phantom and in-vivo in a sheep. The proposed deep learning method outperformed the state-of-the-art algorithm in terms of SNR and paves the way for clinical studies.

The complexity of the MRI protocol is one of the factors limiting the clinical adoption of MR temperature mapping for real-time monitoring of cardiac ablation procedures and a push-button solution would ease its use. Continuous gradient echo golden angle radial acquisition combined with intra-scan motion correction and undersampled temperature determination could be a robust and more user-friendly alternative than the ultrafast GRE-EPI sequence which suffers from sensitivity to magnetic field susceptibility artifacts and requires ECG-gating. The goal of this proof-of-concept work is to establish the temperature uncertainty as well as the spatial and temporal resolutions achievable in an Agar-gel phantom and in vivo using this method. GRE radial golden angle acquisitions were used to monitor RF ablations in a phantom and in vivo in two sheep hearts with different slice orientations. In each case, 2D rigid motion correction based on catheter micro-coil signal, tracking its motion, was performed and its impact on the temperature imaging was assessed. The temperature uncertainty was determined for three spatial resolutions (1 × 1 × 3 mm3, 2 × 2 × 3 mm3, and 3 × 3 × 3 mm3) and three temporal resolutions (0.48, 0.72, and 0.97 s) with undersampling acceleration factors ranging from 2 to 17. The combination of radial golden angle GRE acquisition, simultaneous catheter tracking, intra-scan 2D motion correction, and undersampled thermometry enabled temperature monitoring in the myocardium in vivo during RF ablations with high temporal (< 1 s) and high spatial resolution. The temperature uncertainty ranged from 0.2 ± 0.1 to 1.8 ± 0.2 °C for the various temporal and spatial resolutions and, on average, remained superior to the uncertainty of an EPI acquisition while still allowing clinical monitoring of the RF ablation process. The proposed method is a robust and promising alternative to EPI acquisition to monitor in vivo RF cardiac ablations. Further studies remain required to improve the temperature uncertainty and establish its clinical applicability.

Background Knowledge of the normal myocardial–myocyte orientation could theoretically allow the definition of relevant quantitative biomarkers in clinical routine to diagnose heart pathologies. A whole heart diffusion tensor template representative of the global myofiber organization over species is therefore crucial for comparisons across populations. In this study, we developed a groupwise registration and tractography framework to resolve the global myofiber arrangement of large mammalian sheep hearts. To demonstrate the potential application of the proposed method, a novel description of sub-regions in the intraventricular septum is presented. Methods Three explanted sheep (ovine) hearts (size ~12×8×6 cm3, heart weight ~ 150 g) were perfused with contrast agent and fixative and imaged in a 9.4T magnet. A group-wise registration of high-resolution anatomical and diffusion-weighted images were performed to generate anatomical and diffusion tensor templates. Diffusion tensor metrics (eigenvalues, eigenvectors, fractional anisotropy …) were computed to provide a quantitative and spatially-resolved analysis of cardiac microstructure. Then tractography was performed using deterministic and probabilistic algorithms and used for different purposes: i) Visualization of myofiber architecture, ii) Segmentation of sub-area depicting the same fiber organization, iii) Seeding and Tract Editing. Finally, dissection was performed to confirm the existence of macroscopic structures identified in the diffusion tensor template. Results The template creation takes advantage of high-resolution anatomical and diffusion-weighted images obtained at an isotropic resolution of 150 μm and 600 μm respectively, covering ventricles and atria and providing information on the normal myocardial architecture. The diffusion metric distributions from the template were found close to the one of the individual samples validating the registration procedure. Small new sub-regions exhibiting spatially sharp variations in fiber orientation close to the junctions of the septum and ventricles were identified. Each substructure was defined and represented using streamlines. The existence of a fiber-bundles in the posterior junction was validated by anatomical dissection. A complex structural organization of the anterior junction in comparison to the posterior junction was evidenced by the high-resolution acquisition. Conclusions A new framework combining cardiac template generation and tractography was applied on the whole sheep heart. The framework can be used for anatomical investigation, characterization of microstructure and visualization of myofiber orientation across samples. Finally, a novel description of the ventricular junction in large mammalian sheep hearts was proposed.

Diverse approaches such as oscillating gradients, tensor-valued encoding, and diffusion-relaxation correlation have been used to study microstructure and heterogeneity in healthy and pathological biological tissues. Recently, acquisition schemes with free gradient waveforms exploring both the frequency-dependent and tensorial aspects of the encoding spectrum b(ω) have enabled estimation of nonparametric distributions of frequency-dependent diffusion tensors. These “D(ω)-distributions” allow investigation of restricted diffusion for each distinct component resolved in the diffusion tensor trace, anisotropy, and orientation dimensions. Likewise, multidimensional methods combining longitudinal and transverse relaxation rates, R 1 and R 2, with (ω-independent) D-distributions capitalize on the component resolution offered by the diffusion dimensions to investigate subtle differences in relaxation properties of sub-voxel water populations in the living human brain, for instance nerve fiber bundles with different orientations. By measurements on an ex vivo rat brain, we here demonstrate a “massively multidimensional” diffusion-relaxation correlation protocol joining all the approaches mentioned above. Images acquired as a function of the magnitude, normalized anisotropy, orientation, and frequency content of b(ω), as well as the repetition time and echo time, yield nonparametric D(ω)-R 1-R 2-distributions via a Monte Carlo data inversion algorithm. The obtained per-voxel distributions are converted to parameter maps commonly associated with conventional lower-dimensional methods as well as unique statistical descriptors reporting on the correlations between restriction, anisotropy, and relaxation.

MR thermometry datasets acquired with a GRE radial golden angle sequence at 1.5 T.

1A.I.Virtanen Institute for Molecular Sciences, University of Eastern Finland, Kuopio, Finland. 2Department of Chemistry, Lund University, Lund, Sweden. 3Department of Chemical Engineering, Chalmers University of Technology, Gothenburg, Sweden. 4Department of Medical Radiation Sciences, University of Gothenburg, Gothenburg, Sweden. 5Medical Physics and Biomedical Engineering, Sahlgrenska University Hospital, Gothenburg, Sweden. *email: daniel.topgaard@fkem1.lu.se

Magnetic resonance imaging (MRI) is the method of choice for noninvasive studies of micrometer-scale structures in biological tissues via their effects on the time/frequency-dependent ("restricted") and anisotropic self-diffusion of water. Traditional MRI relies on pulsed magnetic field gradients to encode the signal with information about translational motion in the direction of the gradient, which convolves fundamentally different aspects-such as bulk diffusivity, restriction, anisotropy, and flow-into a single effective observable lacking specificity to distinguish between biologically plausible microstructural scenarios. To overcome this limitation, we introduce a formal analogy between measuring rotational correlation functions and interaction tensor anisotropies in nuclear magnetic resonance (NMR) spectroscopy and investigating translational motion in MRI, which we utilize to convert data acquisition and analysis strategies from NMR of rotational dynamics in macromolecules to MRI of diffusion in biological tissues, yielding model-independent quantitative metrics reporting on relevant microstructural properties with unprecedented specificity. Our model-free approach advances the state-of-the-art in microstructural MRI, thereby enabling new applications to complex multi-component tissues prevalent in both tumors and healthy brain.

We present an algorithm suitable for automatically correcting rolling baseline coming from time-domain truncation induced by the dead time in pulse-acquire one-dimensional MAS NMR spectra. It relies on an iterative estimation of the baseline restricted in the time-domain by the dead time duration combined with a histogram filter allowing adaptive selection of the baseline points. This method does not make any assumption regarding the NMR resonances line shapes or widths and does not modify the acquired free induction decay points. This makes it suitable for accurate deconvolution and quantification of single-pulse MAS NMR spectra. The baseline correction accuracy is evaluated on synthetic solid-state spectra of 19F, 71Ga, and 23Na by comparing the fitted baseline to the theoretical one. The versatility of the algorithm is also exemplified on three additional solid-state spectra of 23Na and 71Ga. The algorithm is made available to the community through a user-friendly standalone Matlab® application.

The increasing recognition of minimally invasive thermal treatment of tumors motivate the development of accurate thermometry approaches for guaranteeing the therapeutic efficacy and safety. Magnetic Resonance Thermometry Imaging (MRTI) is nowadays considered the gold-standard in thermometry for tumor thermal therapy, and assessment of its performances is required for clinical applications. This study evaluates the accuracy of fast MRTI on a synthetic phantom, using dense ultra-short Fiber Bragg Grating (FBG) array, as a reference. Fast MRTI is achieved with a multi-slice gradient-echo echo-planar imaging (GRE-EPI) sequence, allowing monitoring the temperature increase induced with a 980 nm laser source. The temperature distributions measured with 1 mm-spatial resolution with both FBGs and MRTI were compared. The root mean squared error (RMSE) value obtained by comparing temperature profiles showed a maximum error of 1.2 °C. The Bland-Altman analysis revealed a mean of difference of 0.1 °C and limits of agreement 1.5/−1.3 °C. FBG sensors allowed to extensively assess the performances of the GRE-EPI sequence, in addition to the information on the MRTI precision estimated by considering the signal-to-noise ratio of the images (0.4 °C). Overall, the results obtained for the GRE-EPI fully satisfy the accuracy (~2 °C) required for proper temperature monitoring during thermal therapies.