DIMOND: DIffusion Model OptimizatioN with Deep Learning

Advanced Science, Год журнала: 2024, Номер 11(24)

Опубликована: Апрель 18, 2024

Abstract Diffusion magnetic resonance imaging is an important tool for mapping tissue microstructure and structural connectivity non‐invasively in the vivo human brain. Numerous diffusion signal models are proposed to quantify microstructural properties. Nonetheless, accurate estimation of model parameters computationally expensive impeded by image noise. Supervised deep learning‐based approaches exhibit efficiency superior performance but require additional training data may be not generalizable. A new DIffusion Model OptimizatioN framework using physics‐informed self‐supervised Deep learning entitled “DIMOND” address this problem. DIMOND employs a neural network map input optimizes minimizing difference between acquired synthetic generated via parametrized outputs. produces tensor results generalizable across subjects datasets. Moreover, outperforms conventional methods fitting sophisticated including kurtosis NODDI model. Importantly, reduces time from hours minutes, or seconds leveraging transfer learning. In summary, manner, high efficacy, increase practical feasibility adoption clinical neuroscientific applications.

Язык: Английский

---

The Diffusion Exchange Ratio (DEXR): A minimal sampling of diffusion exchange spectroscopy to probe exchange, restriction, and time-dependence

Journal of Magnetic Resonance, Год журнала: 2024, Номер 366, С. 107745 - 107745

Опубликована: Авг. 6, 2024

Язык: Английский

---

Age‐related alterations in human cortical microstructure across the lifespan: Insights from high‐gradient diffusion MRI

Aging Cell, Год журнала: 2024, Номер 23(11)

Опубликована: Авг. 8, 2024

Abstract The human brain undergoes age‐related microstructural alterations across the lifespan. Soma and Neurite Density Imaging (SANDI), a novel biophysical model of diffusion MRI, provides estimates cell body (soma) radius density, neurite density in gray matter. goal this cross‐sectional study was to assess sensitivity high‐gradient MRI toward cortical microstructure adult lifespan using SANDI. Seventy‐two cognitively unimpaired healthy subjects (ages 19–85 years; 40 females) were scanned on 3T Connectome scanner with maximum gradient strength 300mT/m multi‐shell protocol incorporating 8 b ‐values time 19 ms. Intra‐soma signal fraction obtained from SANDI model‐fitting data strongly correlated age all major lobes ( r = −0.69 −0.60, FDR‐ p < 0.001). 0.48–0.63, 0.001) soma 0.28–0.40, 0.04) significantly volume prefrontal cortex, frontal, parietal, temporal lobes. relationship between metrics greater than or comparable regions, particularly occipital lobe anterior cingulate gyrus. In contrast metrics, associations tensor imaging (DTI) kurtosis low moderate. These results suggest that may be more sensitive underlying substrates neurodegeneration aging DTI traditional macroscopic measures such as thickness.

Язык: Английский

---

In vivo Correlation Tensor MRI reveals microscopic kurtosis in the human brain on a clinical 3T scanner

NeuroImage, Год журнала: 2022, Номер 254, С. 119137 - 119137

Опубликована: Март 23, 2022

Diffusion MRI (dMRI) has become one of the most important imaging modalities for noninvasively probing tissue microstructure. Diffusional Kurtosis (DKI) quantifies degree non-Gaussian diffusion, which in turn been shown to increase sensitivity towards, e.g., disease and orientation mapping neural tissue. However, specificity DKI is limited as different sources can contribute total intravoxel diffusional kurtosis, including: variance diffusion tensor magnitudes (Kiso), due anisotropy (Kaniso), microscopic kurtosis (μK) related restricted microstructural disorder, and/or exchange. Interestingly, μK typically ignored signal modelling it assumed be negligible tissues. recently, Correlation Tensor (CTI) based on Double-Diffusion-Encoding (DDE) was introduced source separation, revealing non preclinical imaging. Here, we implemented CTI first time a clinical 3T scanner investigated healthy subjects. A robust framework separation humans introduced, followed by estimation (and other sources) brain. Using this approach, find that significantly contributes both grey white matter but, expected, not ventricles. The maps human brain are presented, spatial distribution provides unique contrast, appearing from isotropic anisotropic counterparts. Moreover, group average templates these have generated time, corroborated our findings at underlying individual-level maps. We further show common practice ignoring assuming multiple Gaussian component approximation introduces significant bias and, perhaps even worse, compromises their interpretation. Finally, twofold acceleration discussed context potential future applications. conclude much vivo characterizations pathological

Язык: Английский

---

Measuring water exchange on a preclinical MRI system using filter exchange and diffusion time dependent kurtosis imaging

Magnetic Resonance in Medicine, Год журнала: 2022, Номер 89(4), С. 1441 - 1455

Опубликована: Ноя. 20, 2022

Purpose Filter exchange imaging (FEXI) and diffusion time ( t )‐dependent kurtosis (DKI( )) are both sensitive to water between tissue compartments. The restrictive effects of microstructure, however, introduce bias the rate obtained by these two methods, as their interpretation conventionally rely on Kärger model barrier limited Gaussian Here, we investigated whether FEXI DKI( ) can provide comparable rates in ex vivo mouse brains. Theory Methods data were acquired from brains a preclinical MRI system. Phase cycling negative slice prewinder gradients used minimize interferences gradients. Results In corpus callosum, apparent (AXR) correlated with (the inverse time, 1/ τ along radial direction. comparison, discrepancies found cortex due low filter efficiency confounding microstructure. Conclusion results suggest that same processes white matter when separated complex microstructure gray matter, potential among multiple compartments still pose challenge for ).

Язык: Английский

---