Browsing by Subject "Diffusion tensor imaging"

There always exists a demand to accelerate the time-consuming MRI acquisition process. Many methods have been proposed to achieve this goal, including deep learning method which appears to be a robust tool compared to conventional methods. While many works have been done to evaluate the performance of neural networks on standard anatomical MR images, few attentions have been paid to accelerating other less conventional MR image acquisitions. This work aims to evaluate the feasibility of neural networks on accelerating Brain DTI and Gynecological Brachytherapy MRI. Three neural networks including U-net, Cascade-net and PD-net were evaluated. Brain DTI data was acquired from public database RIDER NEURO MRI while cervix gynecological MRI data was acquired from Duke University Hospital clinic data. A 25% Cartesian undersampling strategy was applied to all the training and test data. Diffusion weighted images and quantitative functional maps in Brain DTI, T1-spgr and T2 images in GYN studies were reconstructed. The performance of the neural networks was evaluated by quantitatively calculating the similarity between the reconstructed images and the reference images, using the metric Total Relative Error (TRE). Results showed that with the architectures and parameters set in this work, all three neural networks could accelerate Brain DTI and GYN T2 MR imaging. Generally, PD-net slightly outperformed Cascade-net, and they both outperformed U-net with respect to image reconstruction performance. While this was also true for reconstruction of quantitative functional diffusion weighted maps and GYN T1-spgr images, the overall performance of the three neural networks on these two tasks needed further improvement. To be concluded, PD-net is very promising on accelerating T2-weighted-based MR imaging. Future work can focus on adjusting the parameters and architectures of the neural networks to improve the performance on accelerating GYN T1-spgr MR imaging and adopting more robust undersampling strategy such as radial undersampling strategy to further improve the overall acceleration performance.

Understanding the structure of the brain has been a major goal of neuroscience research over the past century, driven in part by the understanding that brain structure closely follows function. Normative brain maps, or atlases, can be used to understand normal brain structure, and to identify structural differences resulting from disease. Recently, diffusion tensor magnetic resonance imaging has emerged as a powerful tool for brain atlasing; however, its utility is hindered by image resolution and signal limitations. These limitations can be overcome by imaging fixed ex-vivo specimens stained with MRI contrast agents, a technique known as diffusion tensor magnetic resonance histology (DT-MRH). DT-MRH represents a unique, quantitative tool for mapping the brain with unprecedented structural detail. This technique has engendered a new generation of 3D, digital brain atlases, capable of representing complex dynamic processes such as neurodevelopment. This dissertation explores the use of DT-MRH for quantitative brain atlasing in an animal model and initial work in the human brain.

Chapter 1 describes the advantages of the DT-MRH technique, and the motivations for generating a quantitative atlas of rat postnatal neurodevelopment. The second chapter covers optimization of the DT-MRH hardware and pulse sequence design for imaging the developing rat brain. Chapter 3 details the acquisition and curation of rat neurodevelopmental atlas data. Chapter 4 describes the creation and implementation of an ontology-based segmentation scheme for tracking changes in the developing brain. Chapters 5 and 6 pertain to analyses of volumetric changes and diffusion tensor parameter changes throughout rat postnatal neurodevelopment, respectively. Together, the first six chapters demonstrate many of the unique and scientifically valuable features of DT-MRH brain atlases in a popular animal model.

The final two chapters are concerned with translating the DT-MRH technique for use in human and non-human primate brain atlasing. Chapter 7 explores the validity of assumptions imposed by DT-MRH in the primate brain. Specifically, it analyzes computer models and experimental data to determine the extent to which intravoxel diffusion complexity exists in the rhesus macaque brain, a close model for the human brain. Finally, Chapter 8 presents conclusions and future directions for DT-MRH brain atlasing, and includes initial work in creating DT-MRH atlases of the human brain. In conclusion, this work demonstrates the utility of a DT-MRH brain atlasing with an atlas of rat postnatal neurodevelopment, and lays the foundation for creating a DT-MRH atlas of the human brain.

Cerebral palsy (CP) is a non-progressive sensory-motor disorder and is one of the most costly and debilitating chronic pediatric conditions, affecting 2-3 out of 1000 children in the United States. To better understand the underlying mechanisms that result in the devastating motor deficits in CP patients, quantitative assessment of brain regions and connectivity with respect to the severity of injury is critical. It is generally known that motor control can be largely influenced by the integrity of the corticospinal tract (CST) connecting the primary motor cortex (pre-central gyrus) to the brainstem and the spinal cord. Thus, we further sought to establish CST volume as a sensitive and reliable biomarker that can reveal functional deficit in individual patients through comprehensive quantitative analyses in 29 consented human subjects (1-6 years of age, mean 2.67±1.36 years) diagnosed with hemiplegic, diplegic, or quadriplegic CP. The proportional reduction of bilateral CST volume with increased disease severity was observed for diplegic and quadriplegic patients (i.e. with bilateral motor deficits). Furthermore, CST volume with respect to disease severity in hemiplegic patients (i.e. with unilateral motor deficits) exhibited more complex patterns of asymmetry, revealing evidence for alternative neuromechanisms of motor control, such as compensation and neuronal plasticity by the unaffected hemisphere. In all cases, other diffusion metrics such as fractional anisotropy (FA) and mean diffusivity (MD) within the CST did not display significant correlation with disease severity. It is thus concluded that individual CST volume, accurately derived from DTI tractography, is one of the strongest non-invasive imaging biomarkers which could potentially help understand the differential causes for motor deficits and compensation in bilateral CP. It could also be further adopted to assess the underlying mechanism for functional recovery in individual patients undergoing cellular therapies. In addition to motor disability, numerous other neurodevelopmental differences are associated with CP and contribute greatly to the disease morbidity. Therefore, we further examined global changes in WM connectivity with respect to CP disease severity with a whole brain connectome analysis in a cohort of 18 pediatric patients with bilateral CP. As expected, reduction in overall and mutual connectivities throughout the motor regions, including the primary and supplemental motor areas, basal ganglia, and brainstem, was associated with the extent of primary motor disability. In addition, our results show diffuse and significant reduction in global inter-regional connectivity, measured as fiber volume, throughout the entire brain in relation to disease severity. This novel finding has not been previously reported in the CP literature and brings CP into line with multiple other pediatric neuropsychiatric disorders that are thought to involve network-level structural and/or functional disruptions early in development.

In recent years, the emergence of diffusion tensor imaging (DTI) has provided a unique means via water diffusional characteristics to investigate the white matter integrity in the human brain, and its impact on neuronal functions. However, since the characterization of white matter integrity using DTI often lacks tissue specificity, most research studies report changes in anisotropy that are not explicitly correlated with particular cellular origins. To improve the utility of DTI in translational neuroimaging, it is critical to develop DTI acquisition techniques that are quantitative and tissue specific.

There are, nevertheless, existing methods for tissue specificity. For example, myelin water images can be generated using multiple echo time (TE) or magnetization transfer techniques. These techniques can detect changes in the concentration of myelin associated markers, but not in their spatial organization. Most white matter pathologies however start with early microstructural changes in the myelin sheaths during which the tissue contents remain similar and are thus not differentiable on a conventional MR image. Thus, the ability to construct a diffusion tensor that is myelin specific can have an immediate impact on our better understanding myelin physiology and pathophysiology during brain development.

Unfortunately, the myelin water signal decays rapidly because of its short transverse relaxation time constant (T2 < 50 ms), especially in DTI experiments where the echo time (TE) can be as large as 100ms. Even in special cases where the TE is shorter, the lack of myelin selectivity in conventional DTI techniques makes assessment of myelin microstructure extremely challenging. Thus we need to develop a DTI methodology that will greatly shorten the TE and allow myelin selectivity.

To address these challenges we have developed innovative DTI acquisition methodologies that can specifically assess myelin microstructural changes in white matter. To preserve more signal from myelin water we used a stimulated echo DTI implementation. In our initial approach we integrated this sequence with a magnetization transfer preparation to achieve additional differentiating sensitization to myelin water and derive a myelin water weighted (MWW) diffusion tensor. Our results indicate that, compared to the conventional DTI, myelin water diffuses along the axis of the fiber, but has the same has larger fractional anisotropy (FA) due to significantly smaller radial diffusivity. The limited specificity of MT and high radio frequency power deposition of MT-DTI restrict its applicability in clinical studies.

To obtain increased myelin specificity we implemented a robust stimulated echo DTI sequence with segmented spiral-out readout trajectory for achieving minimal TE on clinical MRI scanners. To ensure high spatial accuracy throughout the DTI scan we further develop a methodology for inherently and dynamically correcting both motion induced phase errors and off-resonance effects due to magnetic field inhomogeneities (including eddy currents) in the reconstructed image. We the used this technique to conduct an unprecedented experiment in which we collected DTI images at multiple echo times (as short as 18ms) and characterized the dependence of anisotropy on the T2 components including myelin water. The results confirmed the anisotropy characteristics of myelin water found with our initial previous approach.

Building on this new information, we designed a MWW-DTI method based on the simultaneous acquisition of DTI images at two different echo times within clinical practical durations. It is hoped that this new DTI technique sensitized myelin microanatomy will find wide applications in monitoring healthy brain development in pediatric populations, as many developmental brain disorders start with microstructural changes in white matter.