By: Frank Farach, Staff Scientist
Did you know that we may soon be able to control robots with our minds? This was just one of many interesting research applications I learned about at the annual meeting of the Organization for Human Brain Mapping (OHBM) in Seattle last week. OHBM accurately bills itself as the “primary international organization dedicated to using neuroimaging to discover the organization of the human brain.” It was certainly the right place to be to learn the latest in the fields of neuroimaging and neuroinformatics.
The neuroscientists I met had impressive technical skills that were clearly prized by the community. I’ve been to a lot of academic conferences, but this was the first one that planted a large hackathon room right in the middle of the exhibition hall. I witnessed teams of scientists racing to build impressive functional prototypes of cloud-based brain mapping tools in just a few days.
Like many other fields of research, neuroimaging is experiencing data management pains. Raw imaging data must be processed extensively before it can be used to answer meaningful research questions. Many researchers I spoke with said they use highly customized data-processing pipelines, which makes it difficult to compare data processing steps across labs. Happily, open-source projects such as the Neuroinformatics Database (NiDB) are trying to lessen this pain by enabling peer-to-peer sharing of processing pipelines and imaging data. The neuroinformatics community is also building impressive data repositories like the National Database for Autism Research, the Autism Brain Imaging Data Exchange (ABIDE), and the Mind Research Network’s Collaborative Informatics and Neuroimaging Suite (COINS).
Cross-study data integration is another major problem. Because of variability in the way scanners acquire data, differences in instrumentation are a major concern in meta- and mega-analyses. Data sets quickly balloon to many terabytes of binary data files. Researchers in the burgeoning subfield of imaging genetics feel these pains acutely, as they need large samples to examine the relationship between genetic variability and brain structure and function. Once more, it was heartening to see these researchers partner with funding agencies and other organizations to build the tools they need.
Indeed, one of the most exciting themes of the conference for me was that research communities can make great progress on data management problems by collaborating with each other, funding agencies, and other organizations. I look forward to hearing about more collaborative efforts in the neuroimaging community in the year ahead.