Assessment of cerebrovascular permeability by using dynamic susceptibility contrast MRI

Abstract

Cerebrovascular permeability has been considered a standard biomarker for brain tumors or multiple sclerosis that causes severe blood-brain-barrier (BBB) disruption. In addition, as dementia or aging has been reported to damage the BBB, interest in cerebrovascular permeability as a biomarker of degenerative brain disease is increasing. Dynamic contrast enhanced (DCE)-MRI is a standard technique to evaluate cerebrovascular permeability and widely used in clinics. However, DCE-MRI is less sensitive to subtle BBB damaged diseases such as early stage of ischemic stroke and degenerative brain disease, so it sometimes fails to evaluate cerebrovascular permeability. Therefore, improving the sensitivity of cerebrovascular permeability imaging technique is essential. Generally, T2* contrast is more sensitive to MR contrast agent (CA) than T1 contrast. For CA effects on T1, proton should be physically contact to metal ion in CA. However, CA effect on T2* relaxation is based on inhomogeneity of local magnetic field, which means that protons do not need to contact the metal ions of CA. Therefore, T2*-weighted signal may be more sensitive to CA extravasation than T1-weighted signal. DCE-MRI use T1-weighted signal, but there is another MRI technique called dynamic susceptibility contrast (DSC)-MRI that use a T2*-weighted signal In this thesis, firstly, the feasibility of improving sensitivity using DSC-MRI has been studied with temporal middle carotid artery occlusion (tMCAO) rat model. To study the extravasated CA effect on DSC-MRI, two different types of CAs were used, which are Gd-DOTA and monocrystalline iron oxide nanoparticle (MION). MION is one of the blood pool CA and it is too large to leak out of the blood. Therefore, by comparing MION and Gd-DOTA, the extravasated CA effects on DSC-MRI systematically were studied. The results showed that Gd-DOTA injected DSC-MRI of infarction underestimated cerebral blood volume and cerebral blood flow, and contralateral hemisphere did not. Also, the difference between Gd-DOTA and MION of DSC-MRI was strong correlated with the damaged level of BBB. Therefore, DSC-MRI has enough contrast to evaluate vascular permeability for subtle BBB damaged region. Secondly, the sensitivity of DCE-MRI and DSC-MRI was compared with tMCAO model. Also, the risk of incorrect segmentation of normal tissue was reduced through pattern recognition analysis. The results showed that DSC-MRI is more sensitive to CA extravasation than DCE-MRI, and that pattern recognition guided DSC-MRI has advantages in automatically segmenting normal tissue areas without pre-knowledge of the level of BBB damage. Therefore, pattern recognition guided DSC-MRI is suitable for evaluating vascular permeability of subtle BBB damaged region. Lastly, the quantification process of DSC-MRI was improved by using sequentially injected DSC-MRI data. To evaluate vascular permeability, pharmacokinetic model of DSC-MRI forces to define the normal tissue region. However, errors arise due to the ROI dependence and mismatch between the pharmacokinetic model and the DSC-MRI signal. To reduce the errors, vascular permeability was calculated through model-free analysis using additional DSC-MRI data. The results showed that dual DSC-MRI is robust to ROI dependence and reduces errors due to pharmacokinetic models. Therefore, dual DSC-MRI has an advantage in quantifying vascular permeability. In conclusion, vascular permeability of subtle BBB damaged region was successfully evaluated by DSC-MRI and showed that DSC-MRI is sensitive to CA extravasation than DCE-MRI. Although most studies focus on improving sensitivity of DCE-MRI, it is difficult due to the fundamental limitations of T1-weighted signal, and T2*-weighted signal can be a good alternative. The potential of T2*-weighted signal was shown in this thesis but still it is unclear why T2*-weighted signal is more sensitive to CA extravasation than T1-weighted signal. The T2* effects on CA extravasation is so complex that it has not been well studied, but if the T2* effects are deeply understood in further studies, DSC-MRI could be further improved.