Diffusion-encoded MRI for assessment of structure and microcirculation : Aspects of q-space imaging and improved IVIM modelling

Sammanfattning: Diffusion and perfusion MRI are valuable methods for investigating the microstructure and viability of tissue. One pure diffusion study was included in this thesis, with the purpose of studying microstructure and carrying out size estimations with the q-space diffusion imaging method. This is a method that has predominantly been explored with NMR spectrometers. In our study, a biological phantom consisting of asparagus stems was investigated using a clinical MRI unit to gain further knowledge about the q-space methodology in a setting where gradient performance is limited. Even though the q-space method showed limited possibilities, retrieval of some structural information was shown to be feasible.In the remaining three doctoral thesis projects, the intravoxel incoherent motion (IVIM) imaging concept was explored, allowing for extraction of combined diffusion and perfusion information from a given dataset. IVIM imaging is a non-invasive technique for acquiring diffusivity as well as microvascular and perfusion-related parameters using a diffusion-weighted pulse sequence. The model used in this technique is often limited because no relaxation properties are incorporated, and also because the signal component from blood is contaminated by cerebrospinal fluid/free water.One study was dedicated to exploring the IVIM parameters at different field strengths and the influence of relaxation and signal-to-noise ratio (SNR) was observed. Although the repeatability was generally better at higher magnetic field strength, it was shown that the relaxation properties and an unexpectedly low SNR at high field strengths resulted in erroneous blood volume estimates. The model commonly used for IVIM data analysis was then modified to compensate for relaxation effects, based on literature values. When this correction was performed, results from the lower field strengths showed lower discrepancy from expected values, while the results from higher field strength were still erroneous, likely due to physiological noise.In a following project, relaxation times were actually measured during the data collection, and incorporated to compensate for relaxation and improve the fitting procedure. The model was also upgraded to a three-compartment model to better describe the underlying tissue by including the cerebrospinal fluid component. Compared to a non-relaxation-compensated model, the three-compartment model with relaxation-compensated data modified the obtained results and reduced the CSF contamination.The last IVIM project also included a three-compartment model, but in this case the purpose of the third compartment was to improve the quantification of the fraction of free water. The free water fraction has been established as a source of clinically useful image contrast, pointing at pathologies that affect the extracellular space, for example, atrophy and neuroinflammation. With our model, the bias from microcirculation was reduced in the free water estimate. A model where extracellular and microvascular effects can be separated might enable new diagnostic possibilities.