High-resolution multi-parametric molecular imaging via multi-dimensional MR spectroscopic imaging

Wang, Zepeng

Permalink

https://hdl.handle.net/2142/129528 Copy

Description

Title

High-resolution multi-parametric molecular imaging via multi-dimensional MR spectroscopic imaging

Author(s)

Wang, Zepeng

Issue Date

2025-04-25

Director of Research (if dissertation) or Advisor (if thesis)

Lam, Fan

Doctoral Committee Chair(s)

Lam, Fan

Committee Member(s)

• Sutton, Brad
• Anastasio, Mark
• Huesmann, Graham

Department of Study

Bioengineering

Discipline

Bioengineering

Degree Granting Institution

University of Illinois Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• MRI
• MR spectroscopic imaging
• subspace imaging
• subspace-based processing
• multi-parametric mapping
• microstructure imaging

Abstract

Magnetic resonance spectroscopic imaging (MRSI), as an important label-free, noninvasive molecular imaging modality, has demonstrated unique potential in detecting and quantifying molecular/biochemical changes in various disease and neuroscience applications. Nevertheless, besides the inherently limited tradeoffs among SNR, imaging speed and resolution due to the low abundance of imaging targets (i.e., brain metabolites), current application of in vivo MRSI technologies are still limited by (1) insufficient resolution to discern heterogeneous metabolic alterations across different tissue types; (2) insufficient molecular specificity, e.g., measuring a limited set of predominate metabolites thus lacking the specificity to resolve weaker signals such as those from neurotransmitters; (3) insufficient quantitative biomarkers beyond molecule concentrations, e.g., relaxation and diffusion parameters that are commonly measured for water and can provide more specific information to certain pathology or microstructural abnormality in different brain disorders. Therefore, new imaging capabilities for non-invasively mapping high-resolution, metabolite distributions as well as molecule-specific information beyond concentration (e.g., quantitative biophysical properties such as relaxation and diffusion parameters) promise to significantly advance MRSI-based molecular imaging. This thesis research aims to fill these technological and knowledge gaps via developing a new suite of multi-dimensional(MD)-MRSI methods, i.e., introducing additional encoding dimensions (e.g., J-coupling, relaxation, diffusion encoding, dynamic information encoding etc) to each imaging voxel of MRSI. We propose to develop innovative quantitative MD-MRSI techniques to demonstrate unprecedented high-resolution, multi-parametric molecular imaging capabilities. This will not only enhance our understanding of the comprehensive molecular-level underpinning of brain functions and pathophysiological processes associated with neurological disorders but also provide potential new molecular-level biomarkers for improved diagnosis, prognosis and therapeutic monitoring of various neurological and neurodegenerative diseases. Inspired by the subspace imaging concept in SPectriscopic Imaging by exploiting sptiospectral CorrElation (SPICE), the proposed research proposed an augmented subspace modeling framework, which systematically integrates the technical innovations on advanced data acquisition strategies, optimized experimental designs, physics-informed modeling as well as augmented subspace-based spatiospectral processing and parameter estimation methods. Modality-specific adaptations are also considered for different additional encoding conditions. Specifically, we focus on two unique MD-MRSI strategies: (1) J-resolved/multi-TE MRSI, by adding a second spectral/TE dimension to encode the J-coupling evolutions and $T_2$ decays of different molecules, to enable simultaneous brain metabolites, neurotransmitters and their relaxation parameters mapping and (2) DW-MRSI, which incorporates diffusion encoding dimension for mapping molecule-specific diffusion properties, to realize multiplexed, microstructural imaging given the cell-type-specific and predominantly intracellular nature of many metabolites/neurotransmitters. We proposed novel optimized multi-TE MRSI and robust DW-MRSI methods and validated them via numerical simulations, phantom and in vivo experiments. Excellent reproducibility of proposed methods were also demonstrated by test-retest studies. Unprecedented multi-parametric molecular imaging capabilities of the brain was enabled by proposed methods at 3T system. Specifically, metabolite/neurotransmitter mapping and metabolite T2 mapping with highest-ever spatial resolution can be achieved using proposed multi-TE MRSI method within clinical relevant time. Using proposed DW-MRSI methods, high-SNR multi-molecular MD maps, simultaneous metabolite and diffusion coefficient maps, whole-brain metabolite ADC maps can be obtained within clinical relevant time with more than 20-fold resolution improvement compared to standard method. The potential for translating the proposed multi-TE MRSI method to clinical applications was demonstrated via initial investigations on epilepsy patient groups, showing promising preliminary results. This thesis research presented an augmented subspace imaging perspective for MD-MRSI, which not only redefines the state-of-the-art methods for in vivo multi-TE MRSI and DW-MRSI, but also opens up many future research directions. For example, novel acquisition strategies optimized for other additional encoding dimensions can be explored to expand quantitative imaging capabilities of MD-MRSI. New computational tools incorporating sophisticated mathematical models and/or AI-powered methods can be developed to further enhance the performance of MD-MRSI. Moreover, the comprehensive multi-parametric molecular information provided by MD-MRSI may bring new insights into tissue biochemical alterations, potentially advancing its clinical applications across various neurological conditions.

Graduation Semester

2025-05

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/129528

Copyright and License Information

Copyright 2025 Zepeng Wang

Owning Collections