Toward fast and robust in vivo MR quantification of microvasculature

Abstract

Magnetic resonance imaging (MRI) assessments of microvascular anatomy and function in diseases, such as cancer and neurodegenerations are important for detecting abnormal vascular behavior and monitoring therapeutic progress in a noninvasive manner. In MRI, quantitative microvascular biomarkers such as vessel permeability, orientation, blood volume, vessel size index are actively being developed for in vivo applications. Firstly, quantitative vessel permeability information is typically measured by dynamic contrast enhanced (DCE) - MRI, which uses extravasating contrast agent (Gd-DOTA). Following pharmacokinetic modelling is usually applied to dynamical signal change curve after the administration of contrast agent to extract vessel-permeability related parameters. On the other hand, it is generally accepted that there are certain limitations in conventional DCE - MRI acquisitions in terms of its accuracy, and the acquisition speed due to the demanding spatio-temporal tradeoffs for dynamic studies. For example, gradient echo based sequence is typically used for DCE - MRI for high temporal resolution requirements, but induces T2* decay that we often neglect, but becomes significant for high contrast agent concentration regions such as artery or kidney. Tradeoff between spatial and temporal resolution also limits the desired spatial coverage or temporal accuracy of time intensity curves. Secondly, vessel orientation, blood volume, and vessel size index are usually measured by detecting transverse relaxation difference before and after the administration of intravascular T2 contrast agent, such as superparamagnetic iron oxide nanoparticles (SPION). However, transverse relaxation is well known to be affected by unwanted environmental conditions such as air-tissue interface and vessel orientation, which frequently causes severe error in the measurement of blood volume and vessel size index. The subjects and goals of this thesis can be categorized by two sub-sections. In the first section, fast and accurate DCE - MRI was achieved by applying compressed sensing (CS) algorithms, which mitigates the spatio-temporal resolution competition of dynamic acquisitions. Firstly, the optimization for the implementation of compressed sensing to conventional fast low-angle shot (FLASH) sequence which is generally used for DCE - MRI acquisition was performed. After optimization step, temporal or spatial resolution improvements were demonstrated by in vivo experiment, especially in the tumor model. Secondly, compressed sensing was implemented to turbo spin echo (TSE) sequence to minimize transverse artifact by replacing T2* to T2 without reducing temporal resolution and slice coverage. This minimized transverse artifact realized calibration-free T1 estimation from T1-weighted signal intensity. Finally, ultrafast 3D spin echo acquisition was developed by applying compressed sensing to multiple-modulation-multiple-echo (MMME) sequence. Improved enhancement in developed sequence was observed, compared to conventional FLASH sequence with 3D coverage. In the second section, alternative methods to improve accuracy in detecting vessel orientation, blood volume, and vessel size index were developed. Firstly, alternative way to measure blood volume, and vessel size index was suggested and demonstrated by using ultra-short echo time (UTE) sequence. UTE sequence realized the measurement of blood volume with the change of longitudinal relaxation before and after administration of contrast agent, not from that of transverse relaxation. Consequently, accurate blood volume measurement was achieved by longitudinal relaxation which is not sensitive to environmental conditions such as air-tissue interface and vessel orientation. Moreover, alternative vessel size index including longitudinal relaxation showed the potential to reduce the error from environmental conditions. Finally, the new concept of obtaining MR tractography with magnetic field anisotropy was introduced. Compared to the conventional way using susceptibility-induced anisotropic magnetic field inhomogeneity studies, this method doesn’t need re-orientation of the subject utilizing the interference pattern between internal and external field gradients. Developed several methodologies in this thesis for the fast and robust in vivo quantification of microvasculature such as vessel permeability, orientation, blood volume, and vessel size index demonstrated the potentials to improve not only the speed of acquisition but also the accuracy of the in vivo microvascular measurements via efficient sensing and reconstruction MR techniques.