CIAIC Seminar 2024-12-09 Jon-Fredrik Nielsen

Abstract: A magnetic resonance imaging (MRI) experiment is fundamentally very simple: It consists of radiofrequency (RF) and quasi-static magnetic fields that vary in time according to a desired schedule. The hardware to make that happen is ubiquitous and available from several commercial vendors, with nearly identical performance specifications across vendors. Unfortunately, the high degree of harmonization that exists on the hardware side is not matched on the software side: The programming interfaces that allow the sequence designer to control the time-varying magnetic fields are unique to each vendor and difficult to learn, which has (1) resulted in MRI protocols (pulse sequences) that cannot be ported between vendor platforms or even described accurately to a third party, (2) slowed down new pulse sequence development by creating “siloed” developer communities with a high barrier to entry, and (3) made it difficult to know if any observed differences in MRI measurements between sites is due to true biological variation or differences in sequence implementation. Pulseq is both an open file specification for MRI pulse sequences and an associated ecosystem of programming tools and vendor-specific interpreters (drivers) that overcome these barriers. Pulseq has seen significant growth within the pulse sequence developer community in the last 1-2 years, and interpreters for all major MRI vendors are, or will likely soon be, available. In this talk I will describe our role in the Pulseq project including our experience developing a Pulseq interpreter for one of the vendor platforms (GE), and our ongoing efforts to develop and evaluate a Pulseq functional MRI protocol that we hope will enable robust, truly harmonized multi-site and longitudinal functional studies. 

Biography: Dr. Jon-Fredrik Nielsen is Research Associate Professor in the fMRI laboratory at University of Michigan, specializing in MRI sequence design and implementation. He created and manages the Pulseq interpreter for GE scanners, and is co-PI of the HarmonizedMRI project that seeks to implement and disseminate vendor-agnostic functional and quantitative neuroimaging protocols for more reproducible MRI research.