Date of Degree
PhD (Doctor of Philosophy)
Electrical and Computer Engineering
First Committee Member
Magnotta, Vincent A
Second Committee Member
Third Committee Member
Fourth Committee Member
The clinical utility of multi-dimensional MRI applications like multi-parameter mapping and 3D dynamic lung imaging is limited by long acquisition times. Quantification of multiple tissue MRI parameters has been shown to be useful for early detection and diagnosis of various neurological diseases and psychiatric disorders. They also provide useful information about disease progression and treatment efficacy. Dynamic lung imaging enables the diagnosis of abnormalities in respiratory mechanics in dyspnea and regional lung function in pulmonary diseases like chronic obstructive pulmonary disease (COPD), asthma etc. However, the need for acquisition of multiple contrast weighted images as in case of multi-parameter mapping or multiple time points as in case of pulmonary imaging, makes it less applicable in the clinical setting as it increases the scan time considerably. In order to achieve reasonable scan times, there is often tradeoffs between SNR and resolution.
Since, most MRI images are sparse in known transform domain; they can be recovered from fewer samples. Several compressed sensing schemes have been proposed which exploit the sparsity of the signal in pre-determined transform domains (eg. Fourier transform) or exploit the low rank characteristic of the data. However, these methods perform sub-optimally in the presence of inter-frame motion since the pre-determined dictionary does not account for the motion and the rank of the data is considerably higher. These methods rely on two step approach where they first estimate the dictionary from the low resolution data and using these basis functions they estimate the coefficients by fitting the measured data to the signal model.
The main focus of the thesis is accelerating the multi-parameter mapping and 3D dynamic lung imaging applications to achieve desired volume coverage and spatio-temporal resolution. We propose a novel dictionary learning framework called the Blind compressed sensing (BCS) scheme to recover the underlying data from undersampled measurements, in which the underlying signal is represented as a sparse linear combination of basic functions from a learned dictionary. We also provide an efficient implementation using variable splitting technique to reduce the computational complexity by up to 15 fold. In both multi- parameter mapping and 3D dynamic lung imaging, the comparisons of BCS scheme with other schemes indicates superior performance as it provides a richer presentation of the data. The reconstructions from BCS scheme result in high accuracy parameter maps for parameter imaging and diagnostically relevant image series to characterize respiratory mechanics in pulmonary imaging.
Multi-dimensional MRI is a promising imaging modality. Since it is radiation free modality, it has advantages over other imaging modalities like computed tomography (CT), X-rays, etc. MRI acquisition is inherently slow due to hardware limitations that leads to peripheral nerve stimulation. The acquisition speed of MRI limits its diagnostic ability in clinics.
The quantification of multiple tissue parameters from MRI datasets is emerging as a powerful tool for tissue characterization. Parameters such as proton density, longitudinal and transverse relaxation times (denoted by T1 and T2), relaxation times in the rotating frame (T1ρ and T2ρ), as well as diffusion have been shown to be useful in diagnosis of various neurological diseases. Although a single parameter may be sensitive to a number of tissue properties of interest, it may not be specific. Acquiring additional parameters can improve the specificity.
Dynamic lung imaging enables the diagnosis of the abnormalities to the active and passive components involved in respiratory pumping including diaphragm paresis or paralysis, abnormal chest wall mechanics, and muscle weakness, resulting from neuromuscular, pulmonary, or obesity related disorders in dyspnea. It is a powerful non-invasive, non-contrast tool to access abnormalities in regional lung function in pulmonary diseases like chronic obstructive pulmonary disease (COPD), asthma, etc.
However, the need for acquisition of multiple contrast weighted images as in case of multi-parameter mapping or multiple time points as in case of pulmonary imaging increase the scan time considerably. The most common approach is to collect fewer samples. Most MRI images are sparse in some transform domain. In order words they have only a few non-zero coefficients and hence they can be recovered from fewer samples.
Several researchers have developed various schemes for recovering the underlying data with fewer samples. However, they sub-optimal in multi-parameter mapping as well as dynamic pulmonary imaging applications for a number of reasons. Most of these methods make assumptions that are not realistic. For example, some methods assume the data to lie on a low dimensional space which might not true in the event of large interframe motion. Some methods rely on pre-determined information which poorly mimic the actual scenario.
The goal of this thesis is to recover the multi-dimensional MRI data from fewer samples by using a novel scheme called the blind compressed sensing (BCS) scheme. This scheme represents the underlying signal at every pixel as a linear combination of few atoms of the dictionary. The dictionary here is learned from the undersampled data and hence is subject specific. This thesis presents a fast implementation technique to improve the computation speed of the algorithm. The BCS scheme with its fast implementation has enabled several fold acceleration in acquisition without a affecting the image quality. It has the ability to improve some state of the art algorithms used for accelerating multi-dimensional MRI applications.
3D imaging, BCS, Dictionary learning, Lung imaging, MRI, Parameter mapping
xii, 110 pages
Includes bibliographical references (pages 101-110).
Copyright © 2016 Sampada Vasant Bhave
Bhave, Sampada Vasant. "Novel dictionary learning algorithm for accelerating multi-dimensional MRI applications." PhD (Doctor of Philosophy) thesis, University of Iowa, 2016.