Improved MRI Detection of Cerebral Microbleeds with Novel Susceptibility Mapping

脑微出血（CMBs）是小脑出血（MRI小于1cm， CT不可见）。他们的

Cerebral microbleeds (CMBs) are small brain bleeds (less than 1 cm on MRI, and not visible on CT). Their presence is used to make multiple diagnoses (cerebral amyloid angiopathy, traumatic brain injury, acute and chronic hypertensive intracranial hemorrhage, the Alzheimer's drug complication ARIA-H, etc.), and is associated with increased risk of larger brain bleeds, cognitive decline, and aneurysm rupture. They are also associated with worse outcomes for multiple diseases, including brain bleeds in stroke patients after clot busting drugs are given and worse disability after brain clot removal in stroke, relative to patients without CMBs. Detection of CMBs plays a major role in patient therapeutic management based on their number and where in the brain they are found. Unfortunately, the number and location of detectable CMBs are highly variable depending on the type of MRI imaging used. For example, while 3T SWI imaging is the most sensitive of clinically available CMB imaging methods, studies correlating radiology imaging with brain tissue examination have demonstrated that up to 75% of CMBs in tissue are not detected by 3T SWI MRI. Additionally, SWI has been shown to have difficulty distinguishing CMBs from calcification, which is a common CMB mimic. Much of the technical variability seen in SWI and similar clinically used MRI techniques is not present in a newer MRI technique called quantitative susceptibility mapping (QSM). However, most QSM imaging requires the skull to be masked (mQSM), resulting in omission of random parts of the brain in the resultant images. Our work has shown that a newer kind of QSM, that does not require the masking (called pQSM), shows much of the brain not seen on mQSM images. We also showed that CMBs were present in some of the areas of brain that are visible on pQSM but not mQSM images. Typically, the MRIs that diagnose CMBs are read by neuroradiologists. The goal of this project is to study differences in how neuroradiologists read pQSM and SWI images for evaluating CMBs, and to see if those differences would lead to different patient care, related to the number and / or location of CMBs identified. We have access to data from a group of patients admitted to the hospital with larger bleeds, who received MRIs as part of their care that included the raw data needed to make SWI and pQSM images as well as CT scans that show their large brain bleed and areas of calcification. We plan to use this data to make SWI and pQSM images for a panel of reader neuroradiologists to review for CMBs. Because we cannot get brain tissue from these patients, we recruited additional neuroradiologists to serve as an expert panel to select areas on the SWI and pQSM images to classify as CMB or not CMB, to serve as a gold standard. The reader neuroradiologists will then review these areas marked by the expert panel for CMBs.