Your browser has javascript turned off or blocked. This will lead to some parts of our website to not work properly or at all. Turn on javascript for best performance.

The browser you are using is not supported by this website. All versions of Internet Explorer are no longer supported, either by us or Microsoft (read more here:

Please use a modern browser to fully experience our website, such as the newest versions of Edge, Chrome, Firefox or Safari etc.

Probing brain microstructure with multidimensional diffusion MRI: Encoding, interpretation, and the role of exchange


Summary, in English

Diffusion MRI (dMRI) is a non-invasive probe of human brain microstructure. It is a long-standing promise to use dMRI for ‘in vivo histology’ and estimate tissue quantities. However, this faces several challenges. First, the microstructure models used for dMRI data are based on assumptions that may cause erroneous interpretations. Also, probing neurites in gray matter assumes high microscopic diffusion anisotropy in both axons and dendrites, which is not supported by evidence. Furthermore, dMRI data analysis typically ignores diffusional exchange between microscopic environments. This thesis investigates and addresses these challenges using ‘multidimensional’ dMRI techniques that vary additional sequence encoding parameters to obtain new information on the tissue.
In Paper I, we optimized an acquisition protocol for filter exchange imaging (FEXI). We found slow rates of diffusional exchange in normal brain tissue. In patients with gliomas and meningiomas, faster exchange was tentatively associated with higher tumor grade.
In Paper II, we used tensor-valued diffusion encoding to test the NODDI microstructure model. The NODDI assumptions were contradicted by independent data and parameter estimates were found to be biased in normal brain and in gliomas. The CODIVIDE model combined data acquired with different b-tensor shapes to remove NODDI assumptions and reduce the susceptibility to bias.
In Paper III, we used tensor-valued diffusion encoding with multiple echo times to investigate challenges in estimating neurite density. We found that microscopic anisotropy in the brain reflected axons but not dendrites. We could not separate the densities and T2 values of a two-component model in normal brain, but we did detect different component T2 values in white matter lesions. Microstructure models ranked regions from normal brain and white matter lesions inconsistently with respect to neurite density.
In Paper IV, we optimized an acquisition protocol for tensor-valued diffusion encoding with multiple echo times. The data allowed removing all assumptions on diffusion and T2 relaxation from a two-component model. This increased the measurable parameters from two to six and reduced their susceptibility to bias. Data from the normal brain showed different component T2 values and contradicted common model assumptions.
In Paper V, we used tensor-valued diffusion encoding in malformations of cortical development. Lesions that appeared gray matter-like in T1- and T2-weighted contrasts featured white matter-like regions with high microscopic diffusion anisotropy. We interpreted these regions as myelin-poor white matter with a high axonal content. By primarily reflecting axons and not dendrites or myelin, microscopic anisotropy may differentiate tissue where alterations to myelin confound conventional MRI contrasts.
In Paper VI, we used SDE with multiple diffusion times in patients with acute ischemic stroke. Subacute lesions exhibited elevated diffusional exchange that predicted later infarction. MD reduction was partially reversible and did not predict infarction. Diffusional exchange may improve definition of ischemic core and identify additional patients for late revascularization.

Publishing year




Document type



Lund University, Faculty of Science


  • Radiology, Nuclear Medicine and Medical Imaging


  • MRI
  • Diffusion
  • Human
  • Brain
  • Microstructure
  • Fysicumarkivet A:2021:Lampinen



Research group

  • Multidimensional microstructure imaging
  • MR Physics


  • ISBN: 978-91-8039-037-8
  • ISBN: 978-91-8039-038-5

Defence date

28 October 2021

Defence time


Defence place

Föreläsningssal 2, Centralblocket, Entrégatan 7, Skånes Universitetssjukhus, Lund. Join via zoom:


  • Susie Yi Huang (Assistant professor)