<?xml version="1.0" encoding="UTF-8"?><?xml-stylesheet type="text/xsl" href="static/style.xsl"?><OAI-PMH xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/ http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd"><responseDate>2026-11-21T04:13:45Z</responseDate><request verb="GetRecord" identifier="oai:macsphere.mcmaster.ca:11375/31026" metadataPrefix="dim">https://prod-ms-be.lib.mcmaster.ca/server/oai/request</request><GetRecord><record><header><identifier>oai:macsphere.mcmaster.ca:11375/31026</identifier><datestamp>2026-09-24T06:50:12Z</datestamp><setSpec>com_11375_271</setSpec><setSpec>col_11375_272</setSpec></header><metadata><dim:dim xmlns:dim="http://www.dspace.org/xmlns/dspace/dim" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:doc="http://www.lyncode.com/xoai" xsi:schemaLocation="http://www.dspace.org/xmlns/dspace/dim http://www.dspace.org/schema/dim.xsd">
   <dim:field mdschema="dc" element="contributor" qualifier="advisor">Noseworthy, Michael</dim:field>
   <dim:field mdschema="dc" element="contributor" qualifier="author">Tavakkoli, Mitra</dim:field>
   <dim:field mdschema="dc" element="contributor" qualifier="department" lang="en_US">Electrical and Computer Engineering</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="accessioned">2025-02-04T19:18:35Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="available">2025-02-04T19:18:35Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="issued">2025</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="uri">http://hdl.handle.net/11375/31026</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="uri">https://doi.org/10.71548/23574</dim:field>
   <dim:field mdschema="dc" element="description" qualifier="abstract" lang="en_US">Magnetic resonance imaging (MRI) has advanced with the development of accelerated&#xd;
imaging techniques, including parallel imaging, compressed sensing (CS), and more&#xd;
recently, deep learning. These innovations have enhanced MRI speed and image&#xd;
quality, allowing for broader clinical applications. However, adapting these methods&#xd;
to specific organs remains challenging, specially for organs affected by respiratory&#xd;
and cardiac motion, such as the lungs and liver. Lung MRI, for example, suffers from&#xd;
low proton density, limiting the signal-to-noise ratio (SNR) in traditional proton based&#xd;
MR imaging. Hyperpolarized 129Xe imaging provides an effective alternative,&#xd;
enhancing lung visualization but necessitating breath-holds that can be difficult for&#xd;
patients.&#xd;
CS has shown promise in reducing MRI scan durations. In this work, a thorough&#xd;
investigation of its application in hyperpolarized 129Xe ventilation MRI was conducted&#xd;
in Chapter 3. The resultant image quality was found to be sensitive to the chosen&#xd;
Gaussian undersampling pattern, even when the undersampling percentage remained&#xd;
constant. This sensitivity to undersampling patterns, along with parameters like&#xd;
regularization factors, complicates CS optimization and highlights the need for methods&#xd;
with greater flexibility and reduced reliance on manual tuning of reconstruction&#xd;
settings, such as deep learning techniques. In a similar problem, ultra-high-field (7T) MRI holds great potential for achieving higher SNR and improved image quality compared to standard 1.5T and 3T scanners, but image resolution remains limited by breath-hold requirements for motion-sensitive organs. To address this a free-breathing 7T abdominal imaging protocol, employing&#xd;
a pseudo-spiral sampling pattern with CS reconstruction, was developed in Chapter&#xd;
4. This approach demonstrated high image quality while addressing the challenge&#xd;
of patient discomfort associated with breath-holding and reducing motion artifacts&#xd;
through retrospective respiratory gating. However, manually tuning CS to optimize&#xd;
hyperparameters and maximize reconstruction performance can be time-consuming.&#xd;
This suggests that using a deep learning technique with increased freedom could&#xd;
benefit the reconstruction process.&#xd;
Recently, unrolled CS techniques have leveraged the power of deep learning algorithms&#xd;
to overcome the challenges associated with conventional CS. The integration&#xd;
of deep learning into CS facilitates the automatic optimization of underlying parameters,&#xd;
enhancing the quality of reconstructed images. To address the challenges of&#xd;
conventional CS mentioned earlier, the Cascade of Independent Recurrent Inference&#xd;
Machine (CIRIM) was considered a viable alternative in this study. First, unlike CS,&#xd;
which relies on a predetermined l1 constraint to promote image sparsity, CIRIM iteratively&#xd;
enhances image quality through learned priors. Furthermore, previous studies&#xd;
have shown CIRIM's ability to outperform conventional CS and other state-of-the art&#xd;
deep learning (DL) reconstruction techniques. It also demonstrates promising&#xd;
robustness to variations in undersampling patterns, making it an excellent candidate&#xd;
for this study.&#xd;
CIRIM demonstrated its ability to adapt across varying Gaussian undersampling patterns, successfully reconstructing both hyperpolarzied 129Xe lung MRI as discussed&#xd;
in Chapter 5 and 7T breath-hold abdominal data in Chapter 6, outperforming CS&#xd;
in terms of image quality and robustness. Furthermore, CIRIM's adaptability was&#xd;
assessed by training on Cartesian breath-hold abdominal data undersampled with&#xd;
a Gaussian pattern and applying it to prospectively undersampled, free-breathing&#xd;
pseudo-spiral data in Chapter 6. By successfully reconstructing these images, the&#xd;
method demonstrated a particular capacity to generalize across Cartesian and pseudospiral&#xd;
sampling.&#xd;
In conclusion, to achieve high-quality imaging at low scan durations, a robust&#xd;
reconstruction technique capable of handling high acceleration factors is necessary.&#xd;
This study advances the state of the art in MRI by demonstrating that combining&#xd;
the advantages of deep learning reconstruction and non-Cartesian acquisition can address&#xd;
key challenges of conventional CS, including its sensitivity to sampling patterns&#xd;
and the need for manual hyperparameter tuning, while enabling robust, high-quality&#xd;
reconstructions even under suboptimal undersampling conditions. Specifically, this&#xd;
work applies these advancements to 129Xe lung MRI, addressing its sensitivity to&#xd;
Cartesian undersampling, and to free-breathing 7T liver MRI, overcoming challenges&#xd;
associated with breath-hold requirements, motion artifacts, and scan efficiency. Future&#xd;
work should build on these advancements by evaluating alternative non-Cartesian&#xd;
sampling methods, such as radial sampling, to further optimize imaging for motionsensitive&#xd;
organs. By tailoring sampling strategies and improving reconstruction robustness,&#xd;
these techniques can enhance imaging resolution, increase scan efficiency,&#xd;
and address patient comfort through tolerable scan times, paving the way for broader&#xd;
clinical and research applications.</dim:field>
   <dim:field mdschema="dc" element="description" qualifier="degreetype" lang="en_US">Dissertation</dim:field>
   <dim:field mdschema="dc" element="description" qualifier="degree" lang="en_US">Doctor of Philosophy (PhD)</dim:field>
   <dim:field mdschema="dc" element="description" qualifier="layabstract" lang="en_US">Recent advancements in magnetic resonance imaging (MRI) have improved imaging speed and image quality. However, despite its diagnostic power, a limitation of MRI is the time required to acquire data. Traditionally, faster acquisition came at the expense of resolution or signal-to-noise ratio (SNR). In the 1990s parallelizing the acquisition with multiple receivers reduced the acquisition time but incurred an SNR penalty.  &#xd;
Compressed sensing (CS) has since gained attention for faster MRI but is constrained by its sensitivity to manual tuning of the regularization parameter and the choice of Cartesian undersampling patterns. &#xd;
In this work, the sensitivity of CS to Cartesian undersampling patterns in the application of 129Xe ventilation MRI was first thoroughly investigated in Chapter 3, and then deep learning reconstruction demonstrated potential in partially addressing these limitations in Chapter 5. Additionally, a free-breathing abdominal MRI protocol using pseudo-spiral sampling patterns was developed at ultra-high-field MRI (7T), employing CS in Chapter 4 and deep learning-based techniques in Chapter 6. These advancements contribute to making MRI more robust and adaptable for the challenging clinical applications mentioned.</dim:field>
   <dim:field mdschema="dc" element="language" qualifier="iso" lang="en_US">en</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en_US">MRI, Fast Imaging, Lung, Liver, Compressed Sensing, Deep Learning</dim:field>
   <dim:field mdschema="dc" element="title" lang="en_US">ADVANCING LUNG AND LIVER MRI: FROM COMPRESSED SENSING TO DEEP LEARNING</dim:field>
   <dim:field mdschema="dc" element="type" lang="en_US">Thesis</dim:field>
   <dim:field mdschema="others" element="access-status">open.access</dim:field>
</dim:dim></metadata></record></GetRecord></OAI-PMH>