Investigating Quantitative Susceptibility Mapping for Preclinical Cerebral Cavernous Malformations at High-Field MRI

Cerebral Cavernous Malformations (CCM) is an abnormal feature of microvasculature that occurs in the central nervous system. More specifically, CCM is a genetic condition caused by somatic loss-of-function mutations of the three genes KRIT1, CCM2, or PDCD10. Although the mechanism behind the three genes is not yet fully understood, studies have shown that these genes have an important role in the formation of enlarged, thin-walled micro-vessel lesions that are prone to bleeding. This condition is predicted to afflict 0.4-0.8% of the general population. Current methods of treating CCM lesions are microsurgical resection, stereotactic radiosurgery, and conservative management. However, determining the ideal treatment strategy requires further understanding of the patient’s condition and the risks associated with each lesion. To further advance the knowledge of this genetic condition and to enable further research into future treatment candidates, a murine model of CCM that is caused by the somatic loss of function of the same genes was established. These mice possess longer survivability and higher lesion loads that are easily visible on T2-weighted MRIs. The Petr Tvrdik lab at UVA has developed this model with the objective of further characterizing the malformation development on a vascular level and the influence of therapeutic interventions, including targeted therapies using focused ultrasound blood brain barrier opening. To assess lesion development, traditional Magnetic Resonance Imaging (MRI) methods are used. MRI has proven to be the most effective imaging modality for detecting and characterizing CCM lesions in clinical practice. The lesions have a heterogenous signal on MRI which results from an accumulation of hemoglobin degradation products. Previous works by our research team have shown T2-weighted MRIs of in vivo models exhibiting a hypointense rim caused by hemosiderin deposits from recurring microhemorrhages. Although these traditional MRI methods produce useful results, there is a lack of research in cumulative hemosiderin deposit and lesion growth. This thesis aims to enhance comprehensive CCM lesion characterization by (1) establishing the use of Quantitative Susceptibility Mapping (QSM) in our mouse model, (2) determining a suitable QSM method for preclinical high-field MRI, (3) determine whether QSM has quantitative capabilities at 9.4T.
Two experiments were completed to address our objectives. In the initial experiment, the efficacy of QSM was evaluated by employing popular QSM toolboxes with default parameters. The outcomes revealed consistency of magnetic susceptibilities with expectations drawn from existing literature regarding the magnetic susceptibility of our wild-type mice. Notably, CCM lesions in our murine models were discernable as hyperintense bright spots in the QSM analysis. However, all methods exhibited varying but notable negative susceptibility and staircasing artifacts at the lesion sites, presenting an area for refinement.
The subsequent experiment aimed to validate the quantitative robustness of QSM by employing an iron (III) chloride (FeCl3) phantom. Multiple vials, each containing FeCl3 at linearly increasing concentrations, were prepared, scanned, and assessed for magnetic susceptibilities using the same QSM inversion methods and hyperparameters applied in the initial experiment. This investigation was crucial in validating the quantitative accuracy of QSM across a spectrum of known iron concentrations, providing valuable insights into its reliability and potential variations in susceptibility measurements. All inversion methods except NDI proved to have a strong linear correlation between iron concentration and iron magnetic susceptibility. Comparing our results to literature, Morphological Enabled Dipole Inversion (MEDI) proved to be the most accurate method for measuring iron concentration using magnetic susceptibility values.