Please use this identifier to cite or link to this item:
http://cimat.repositorioinstitucional.mx/jspui/handle/1008/1035
CHARACTERIZING THE ERROR OF THE GAUSSIAN PROFILE TO MODEL AXON BUNDLES DIFFUSIVITY | |
ALFONSO ROJAS OAXACA | |
Acceso Abierto | |
Atribución-NoComercial | |
MATEMÁTICAS INDUSTRIALES | |
Diffusion Tensor Imaging (DTI) is one of the most widespread techniques for obtaining information about the underlying tissue geometry and integrity in brain white matter. Despite its limitations and strong assumptions, it is a robust model, widely used in clinical applications. Some examples include the detection of Ischemia; the study of alteration produced by conditions, such as Multiple Sclerosis and Alzheimer’s. Two important limitations of DTI are the impossibility of describing diffusion in substrates with separations and compartments, and the high count of degrees of freedom. Nevertheless, the model can be expanded to represent the signal as the sum of contributions from different tensors; the multi tensor model. Also, some constraints can be assumed to decrease the number of free parameters. In this thesis work, we test constrained multi tensor models (i.e Zeppelin and Stick) through Montecarlo simulations in order to gain some insights on the limits of assuming Gaussian diffusion in the description of neuroanatomy. Montecarlo simulation of the diffusion process is performed in several substrates, which resemble white matter microstructure. Physical parameters are obtained by fitting multi tensor models in one fiber bundle, and crossings of two bundles. The MRI signal obtained by the simulator is free of noise and artifacts to factor out any other source of variability, besides the stochastic simulation itself. Finally, the fitted parameters of the model are compared to the statistics of the spin trajectories of the simulator. We observe low errors in the case of extracellular compartment diffusion. In the case of a single complete bundle, simple Zeppelin diffusion underestimation worsens as the volume fraction increases. Zeppelin + Stick model improves quality substantially, with deviation also increasing proportional to volume fraction. Rotation angle does not seem to play an important role in the estimation error. In the crossing case, the use of multiple shells reduces error substantially. | |
08-11-2019 | |
Tesis de maestría | |
OTRAS | |
Versión aceptada | |
acceptedVersion - Versión aceptada | |
Appears in Collections: | Tesis del CIMAT |
Upload archives
File | Description | Size | Format | |
---|---|---|---|---|
TE 760 (2).pdf | 2.62 MB | Adobe PDF | View/Open |