Using Advanced Multi-Echo NODDI to Enhance Understanding of Stroke-Related Microstructural Dynamics
Ezequiel Farrher, Kuan-Hung Cho, Chia-Wen Chiang, Ming-Jye Chen, Sheng-Min Huang, Li-Wei Kuo, Chang-Hoon Choi, N. Jon Shah
24th July 2025
This research investigates how accurate mathematical modelling of the diffusion- and transverse-relaxation-weighted MRI signal can enhance our understanding of the cascade of biophysical mechanisms triggered following the onset of ischaemic stroke in brain tissue.
Ischaemic stroke triggers a cascade of biological changes in brain tissue, including cellular swelling, axon beading, and microstructural disintegration. One of the primary non-invasive tools for studying these processes in vivo is diffusion MRI, which, via the measurement of the water diffusion properties, provides access to tissue structural properties at the mesoscopic length-scale, i.e., well below the image resolution. During the last decade, the neurite orientation dispersion and density imaging (NODDI) model has been extensively applied to extract compartment-specific information about neurite (i.e. axons and dendrites) density and orientation. However, conventional NODDI is limited by its sensitivity to echo time (TE): compartmental signal fractions are affected by differences in transverse relaxation (T2) across tissue compartments, potentially biasing the estimation of microstructural parameters. The multi-echo NODDI (MTE-NODDI) model was developed to address this limitation by incorporating multiple echo times, allowing separation of T2 effects from diffusion-driven signal components. Nonetheless, MTE-NODDI retains a simplifying assumption of a fixed, brain-wide intrinsic diffusivity (d), which is unrealistic in pathological tissue such as ischaemic lesions.
In this study, the authors evaluated the applicability of an improved estimation approach of MTE-NODDI in a rat model of middle cerebral artery occlusion (MCAo). In this novel version, a modified parameter estimation strategy in which intrinsic diffusivity is released was investigated via computer simulations and in experimental MRI data, while measures were taken to mitigate parameter degeneracy and instability.
This approach enables a more accurate quantification of diffusion and T2 properties, not only within ischaemic tissue, but also in healthy brain tissue. Using the technique, the study monitored the spatiotemporal evolution of tissue properties post-stroke. The findings indicate a significant reduction in intrinsic diffusivity in the ischaemic core, which propagated effects across other MTE-NODDI parameters. Notably, the fraction of signal arising from free fluid increased substantially in the affected tissue, diverging from previous findings obtained with conventional MRI. Measures of water proton relaxation (T2 values) inside and outside neurites also showed marked increases, reflecting changes in tissue water content, microstructure and chemical composition.
More generally, the results demonstrated heterogeneous dynamics between the ischaemic core and surrounding peri-infarct tissue, underscoring distinct patterns of pathological progression across regions. Parameters such as free-water fraction, water mobility, and compartment-specific relaxation times followed different temporal trajectories, highlighting the complexity of tissue remodelling following ischaemia. These advanced quantitative measures enable a more detailed and mechanistic understanding of the biophysical processes underlying stroke.
Overall, this work represents a substantive advance in quantitative MRI of stroke, providing a framework for accurate, compartment-specific characterisation of tissue microstructure in vivo. The methodology offers critical insights that can inform both preclinical research and the interpretation of clinical imaging, ultimately supporting the development of improved diagnostic and therapeutic strategies.
Origional publication: On the use of multi-echo NODDI MRI with released intrinsic diffusivity for the assessment of tissue diffusion and relaxation properties in experimental ischaemic stroke