NINDS Quantitative MRI Core Facility

National Institute Of Neurological Disorders And Stroke

Investigators

Linked publications, trials & patents

Abstract

- Brain segmentation algorithms: Neurological diseases can alter the brain volumes, and the core has focused on robust automated brain segmentation and volumetric methods. Core personnel have previously developed a machine learning based atlas-free brain segmentation technique for clinical 3T (Classification using DErivative-based Features or C-DEF) and a transfer-learning based method (Pseudo-Label Assisted nnU-Net or PLAn) for segmenting brain scans acquired on 7T scanners. We are currently applying these techniques to several neurological disorders such as multiple sclerosis (MS), cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), and HIV (including clinical trial on safety of drug pembrolizumab in HIV-positive people). In addition, conventional brain segmentation algorithms such as FreeSurfer are used for finer parcellations in studies such as long-term effects in the brain of people who have recovered from COVID-19 infections. Major equipment needed for these techniques include GPU servers, and 50% effort from a professional programmer is needed to maintain/improve the techniques and train users. (Reference #4, 6, 7, 8 in bibliography) - Imaging of postmortem brain tissue: We have previously developed high-resolution imaging techniques for the postmortem fixed brain. The core also has expertise to implement MRI-guided histopathology utilizing sample-specific or individualized 3D-printed cutting boxes. These are widely used in collaborative projects, such as identifying regions of brain metastasis in small-cell lung cancer, identifying pathological features in various brain regions of progressive multifocal leukoencephalopathy (PML), and locating lesions in MS. The core is also working on collaborative projects to perform MR-Microscopy (imaging at finer than 100-micron isotropic resolution) in various regions of the human brain to better understand the cytoarchitectural contributions to MRI contrast. Finally, we are preparing to extend the postmortem imaging techniques to study the spinal cord in various neurological disorders. Major equipment needed for this technique is custom designed tissue chambers, recirculating chiller, portable freezer, Fomblin, and 3D printer. (Reference #5 in bibliography) - In-vivo spinal cord imaging: The core has previously developed high resolution spinal cord imaging techniques at 3T and 7T. The analysis software was packaged for distribution to collaborators, and the atrophy imaging sequence is being used by various collaborators within and outside the NINDS. We have also started exploring the use of spinal cord coil in patients with diseases such as MS. - In-vivo imaging at ultra-low fields: Core is involved closely with developing and applying imaging techniques on the ultra-low field MRI scanners, including better visualization of MRI contrast agent (CA) and improving imaging resolution of routine sequences. CA extravasation in the brain is important to increase the clinical utility of mobile and economical ultra-low field MRI scanners. The core has been involved in determining the relaxation properties of various CAs that could be used in clinical practice at 64mT. Based on these relaxation properties, we have designed and successfully implemented a pulse sequence to visualize contrast enhancement in white matter lesions of MS and PML patients. We hope to improve this approach and extend it to image CA extravasation into CSF, as well as to study newly developed CAs that are expected to be more effective at ultra-low fields.