Development of the Connectome II MRI Scanner

Eunice Kennedy Shriver National Institute Of Child Health & Human Development

Investigators

Linked publications, trials & patents

Abstract

We are continuing to develop novel diffusion MRI-based pulse sequence, acquisition, and signal modeling frameworks to enable us to "see" or detect fine-scale structures in the living human brain that are three orders of magnitude smaller than the underlying MRI voxel size, and which are currently invisible using clinical MRI scanners. To be concrete, we acquire images with isotropic voxels that are about 1.5 mm on each side, but attempt to observe features of microscopic objects, such as axon diameters and axon diameter distributions, etc., which require a spatial resolution of about 1-2 microns. One way we accomplish this feat is to develop advanced mathematical/physical models describing the relationship between the observed MR signal and various microstructural parameters. It is also important to correct for various artifacts that can blur or corrupt these images, leading to incorrect estimates of imaging quantities. Then, we attempt to infer the biophysical basis of these signals. One method to translate to the Connectome 2.0 is AxCaliber MRI, an approach we invented and developed at the NIH, but which was limited in its resolution on conventional MRI scanners. The new Connectome 2.0 scanner allows one to detect axon diameter distributions with greater precision and accuracy. Another approach we are migrating is mean apparent propagator (MAP) MRI, a method that measures the net displacement distribution or average propagator of diffusing water molecules in tissue. This provides information about the different microenvironments water finds itself in within living brain tissue. Another approach we translating to the Connectome scanner is Time-Scaling MRI, which entails obtaining Mean Apparent Propagator (MAP) MRI data at different diffusion times. This approach allows us to infer certain features of hierarchical tissue organization, such as the fractal dimension that we can exploit to provide mesoscopic and microscale information. We also are investigating various multiple-pulsed field gradient (mPFG) MRI methods for clinical translation, some of which we have previously developed in our lab, which we are extensively vetting, and working to migrate to this powerful new clinical scanning platform. A new mPFG methodology we have pioneered is diffusion tensor distribution (DTD) MRI, which we use to study the heterogeneity of water diffusion within a voxel. This effort has been greatly enhanced in terms of experimental design, computational speed of processing, and by shoring up mathematical underpinnings. We have also been developing macaque and marmoset brain atlases, which allow histological image data to be merged and compared with MRI data of the same areas to enable us to test and vet various MRI methods we develop. In the coming years, much additional vetting and testing will be required to ensure the accuracy and precision of our acquisition and modeling pipelines so that they are ready for clinical implementation and testing in the "out years" of this grant, now that the prototype Connectome 2.0 scanner has been delivered and is ready for scanning normal volunteers and clinical subjects.