A harmonized vendor-agnostic environment for multi-site functional MRI studies

University Of Michigan At Ann Arbor, Ann Arbor MI

Investigators

Linked publications & trials

Abstract

Since its invention in the early 90s, functional magnetic resonance imaging (fMRI) has revolutionized our un- derstanding of the human brain. Functional MRI may be used to observe brain function during a specific motor or cognitive task, or at “rest” (resting-state fMRI). The latter produces so-called “functional connectivity” maps that may provide a new window into human cognition. There is currently a large, world-wide effort underway to discover potential research and clinical uses of such connectivity maps. A significant practical barrier in this effort, however, is the difficulty in ensuring that fMRI experiments are conducted in a consistent and reproducible manner across different centers. In particular, it is generally not possible to ensure identical execution of MR measurements (pulse sequences) across sites operating different MR scanners. Furthermore, even the image reconstruction and data processing methods can be difficult to harmonize, particularly across different MR ven- dors. This makes it challenging to directly compare results between sites, or “pool” data from multiple sites to increase statistical power and gain access to rare clinical conditions. We will assemble and disseminate a truly harmonized, cross-vendor, and flexible environment for fMRI re- search that ensures consistent data acquisition and image reconstruction across sites. Our framework is based on an open-source MR sequence development platform that allows any arbitrary MR pulse sequence to be de- signed “off-line” in Matlab or Python and exported to a vendor-independent file format, that can be ported directly to scanners from different manufacturers (at present, General Electric and Siemens are supported, but others may follow in the future). Due to this open pulse sequence structure it will also be possible to compose a unified image reconstruction environment based on current open-source libraries. Based on this technology, we will provide the fMRI research community with a complete and portable workflow for fMRI data acquisition and image reconstruction, backed up by integrated quality control procedures.