TimVanMourik
commented
5 years ago Hi @jeromedockes, I’m happy to tell you that we’d like to host your lightning talk in the OSR in the neuroscience toolkit session. This will be a talk of 5 minutes + 5 minutes of questions. We’ll update the program in the ReadMe.md shortly. We’d much appreciate it if you could submit slides and other presentation material to the [presentations folder](https://github.com/ohbm/OpenScienceRoom2019/tree/master/ by means of a Pull Request to this repository, preferably but not necessarily before the presentation.
jeromedockes
emdupre
Title
Neuroquery: mapping text to brain regions.
Presentor and Affiliation
Jérôme Dockès, INRIA (@jeromedockes)
Collaborators
Russel Poldrack, Stanford University Fabian Suchanek, Telecom Paris Bertrand Thirion, INRIA (@bthirion) Gaël Varoquaux, INRIA (@GaelVaroquaux)
Github Link (if applicable)
Python package to be released on Github in the near future; but the tool is already online: https://neuroquery.saclay.inria.fr
Abstract (max. 200 words):
Neuroquery is a new tool for large-scale meta-analysis based on multivariate regression and distributional semantics. It is inspired by Neurosynth, but presents key differences:
It expands queries to related terms. Existing meta-analysis methods rely on a fixed vocabulary and can only produce brain maps for a restricted set of expressions. Neuroquery relies on co-occurrences in the literature to automatically map user queries to the set of terms it can encode into brain images. It can thus produce maps for a much wider and more flexible set of queries, greatly reducing the need for users to manually adjust their questions to the tool's vocabulary.
It can encode rare expressions (e.g "prosopagnosia") that are challenging for traditional meta-analysis methods.
Rather than single words, it can encode text of arbitrary length. It can produce meaningful maps for concepts that are better described by a sentence or paragraph, or even for abstracts or papers.
It models the similarities and relationships between terms used in neuroscience publications, allowing users to easily navigate from a query to related terms and documents and comparing their associated brain maps.
Preferred Session
Lightning talk, preferably: "1. Neuroscience toolkit" or "2. Multi-modal research".
Additional Context
A more comprehensive description of the methods behind Neuroquery, and extensive empirical validation, will be presented during the oral session "Modeling and Analysis Methods - Uni/multi-variate analysis", on Monday, June 10:
4554 - Towards Universal Brain Encoding with Multivariate Regression and Large Scientific Corpora
At the Open Science Room, I would like to show the online tool and its features, and receive questions and suggestions from potential users and contributors.