Date of Award

2-1-2022

Embargo Period

4-22-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Neuroscience

College

College of Graduate Studies

First Advisor

Jens H. Jensen

Second Advisor

Dorothea D. Jenkins

Third Advisor

Joseph A. Helpern

Fourth Advisor

Andreana Benitez

Fifth Advisor

Mark A. Eckert

Abstract

Compared to histology, few methods can provide such elegant insight into brain tissue microstructure. While still considered the gold standard for microstructural validation, as it has micron-level resolution, it is time-consuming and typically performed postmortem which limits its practicality. Non-invasive diffusion MRI (dMRI) is sensitive to the diffusion of water which is typically within the micrometer length scale. Clinically, dMRI is used widely to visualize ischemic lesions after stroke in a qualitative sense. As many structures in the brain tissue milieu are on this micron-level order of magnitude, such as axons that have diameters ~1–10 microns, dMRI is well-situated to probe tissue microstructure. Even still, the biophysical interpretation of the dMRI signal in healthy, let alone injured, tissue is complex, and validation is still ongoing. In clinical research, there are many dMRI applications, but for this dissertation we studied neonatal brain development after either preterm or hypoxic-ischemic (HI) birth using dMRI. Typically, the eventual motor and cognitive deficits that arise due to these insults do not show up until later in development, around 12 months. Our objective was to determine if dMRI, specifically diffusional kurtosis imaging (DKI), acquired early in development, near day of life 5–7, could predict long-term outcome scores as measured by standardized developmental testing at 12–18 months of age. Additionally, we also sought to determine the effect, if any, that novel treatment regiments for both preterm infants receiving novel non-invasive brain stimulation paired with bottle feeding and term aged HI infants getting N-acetylcysteine would have on the quantitative diffusion metrics we can derive from DKI, namely the diffusivity and kurtosis. Lastly, we wanted to see if any of the derived DKI metrics could help identify responders to the novel stimulation therapy or provide indicate HI severity. If shown to be helpful, advanced quantitative dMRI would help elicit inform the initiation of earlier and more targeted habilitation therapies for this critical population of neonates. As we will show in this dissertation, there was marked improvement into long-term outcome model predictions using kurtosis metrics. Moreover, it was found that the mean kurtosis was able to delineate those who would respond or not to the novel stimulation therapy in preterm feeding. It was also found that both the fractional anisotropy and radial kurtosis increased more from pre- to post-treatment in responders than in non-responders. This would indicate a greater degree of restriction and directionality to the water movement and thus greater cellular integrity of axons and/or cellular barriers. Lastly, we found that there exist differences in the kurtosis based on the timing for the scan in acute (decreased kurtosis) or convalescing (increased kurtosis) HI. Interestingly, we also found that the kurtosis fractional anisotropy, given the degree of anisotropy of the kurtosis tensor only, provided support in modeling HIE severity and outcomes. DKI has not widely been studied in neonatal development, with only a handful of studies published. This work represents another building block to fully characterize this method which improves upon the current clinical standard diffusion tensor imaging (DTI) in identifying treatment response and, insult severity stage, and helping predict long-term outcomes. This dissertation also focused on the development of more advanced dMRI methods than are typically employed or ready for translation to the clinical setting. The main technique we developed and characterize in depth throughout this body of work is fiber ball imaging (FBI). One of main features of FBI compared to DKI is the use of large diffusion weightings or b-values. DKI typically uses b-values of 1000 and 2000 s/mm2 whereas FBI requires at least 4000 s/mm2 and many sampling directions. The large diffusion weighting isolates the intra-axonal water by acting as a filter for extra-axonal water that is a confounding element at the lower b-values of DKI. With FBI, we can estimate what is known as the fiber orientation density function (fODF) which is related to the angular spread of axon fiber bundles. Using the fODF, and its associated elements, we can calculate the fractional anisotropy of axons (FAA) and the Matusita anisotropy (MAA) that provide insight into the intra-axonal white matter (WM) space specifically. Furthermore, we developed a method to quickly and easily rectify negative values that arise during the fODF estimation. We also show that this method can be used for any fODF estimation technique as well. Using a novel diffusion protocol, triple diffusion encoding (TDE), we provide a means to get a more reliable value for the intrinsic intra-axonal diffusivity, which can slightly improve the fODF estimation. TDE is simple to implement and has only a few modeling assumptions, but as it involves no numerical fitting it can serve as a basis for comparison and validation of various single diffusion encoding white matter models whose data is easier to acquire but require more modeling assumptions. Lastly, we have devised a way to use the fODF as the basis for comparisons rather than its typical use as an input for other models or white matter fiber tractography. We call this high-fidelity FBI. Briefly, we rotate each fODF (there is one for each white matter imaging voxel) into a local coordinate frame such that they are now able to be compared within and across subjects. This is done using the Matusita distance which gives a single value relating to the similarity or difference of the fODFs being compared. As the fODF is a physical entity, we believe that subtle changes in the fine structure of the fODF, say across age, will provide a new dMRI contrast to observe changes in white matter pathology or aging and dementia, such as Alzheimer’s disease. In total, this thesis work has so far led to 5 peer-reviewed papers, with 2 more planned, and several abstracts at international conferences.

Rights

All rights reserved. Copyright is held by the author.

Download

Share

COinS