A statistical re-examination of the spoiled gradient recall echo equation in the analysis of data from DSC and DCE MR imaging procedures
Loading...
Files
Date
2025
Authors
Hu, Yuankun
Journal Title
Journal ISSN
Volume Title
Publisher
University College Cork
Published Version
Abstract
Magnetic Resonance Imaging (MRI) is a non-invasive technique that has become an essential tool in diagnosing and staging a wide range of acute and chronic illnesses, including brain cancer. Motivated by a publicly available ACRIN-6684 dataset from an MRI brain tumour imaging study, this thesis investigates alternative approaches for analysing Dynamic Susceptibility Contrast (DSC) and Dynamic Contrast Enhanced (DCE) data, to recover local perfusion and retention characteristics of the administered paramagnetic contrast agent. This approach, named as direct optimisation of the extended Tofts and Kermode (ETK) model, enables effective conversion of MR signal intensity to contrast agent concentration across various anatomical tissues, while mitigating associated artifacts. The direct optimisation of the ETK model method is based on optimisation frameworks that eliminate the need for pre-scan quantification. It simultaneously estimates the contrast agent concentration, the baseline longitudinal relaxation time constant, and relevant perfusion parameters.
The unique recovery of parameters in the proposed method is established through identifiability analysis, and its performance is validated through both numerical simulations and application to real clinical datasets. Additionally, the root mean square error (RMSE) of the reconstructed signal intensity using the direct optimisation of the ETK model is consistently lower than that achieved by traditional conversion techniques, suggesting improved accuracy and robustness. The variations in signal reconstruction may contribute to discrepancies in the estimation of perfusion parameters across both DSC and DCE imaging protocols. As a result, the enhanced accuracy provided by the direct optimisation of the ETK model has the potential to yield more reliable concentration estimation and, consequently, more accurate perfusion quantification.
Description
Partial Restriction
Keywords
MRI , Concentration reconstructing , Kinetic analysis
Citation
Hu, Y. 2025. A statistical re-examination of the spoiled gradient recall echo equation in the analysis of data from DSC and DCE MR imaging procedures. PhD Thesis, University College Cork.