Failure to load association maps by coordinate in Neurosynth

Has anyone else had issues with Neurosynth not loading terms (from the “Associations” tab) when you give it a coordinate location? When I give it a coordinate location, I am able to see the functional connectivity map and the list of studies that have a peak near that location, but I’m no longer able to see terms. (I’ve seen this happen in the past for some locations, but now it seems to happen regardless of the location.)

This was a particularly useful function of Neurosynth, so I hope it can be recovered! Or, alternatively, are other atlases that people would recommend to serve the same function, i.e., putting in a coordinate location in MNI space and getting back a quick summary of what concepts tend to be associated with that location? The FSL Harvard-Oxford atlas is sometimes useful for this, but Neurosynth was much richer than what’s provided there…

Hey,

Sorry about that. Neurosynth is currently in a state of minimal maintenance and it’s a minor miracle it still works at all.

The goal is to replace its functionality with user-defined meta-analytic workflows as part of the Neurosynth Compose project.

That said, I am the maintainer as such, and I can take a look.

In the meantime, for anything that is broken in Neurosynth, please take a look at NiMARE. It may be a bit more work but it should ultimately support all of the core functionality of the Neurosynth web app.

Best,
Alejandro

@michaelcohen, I think you can do the following:

  1. Create a Nilearn masker object with the spherical ROI around the coordinate (6 mm radius).
  2. Load the Neurosynth dataset into a NiMARE Dataset object (there are instructions for that in the NiMARE documentation).
  3. Use the ROI image and the Dataset object with NiMARE’s ROIAssociationDecoder.

Thanks, it’s good to know that there is alternative approach via NiMARE (I wasn’t familiar with this before)!

I do hope you guys are able to get the Web site issue fixed though, as it’s so nice, when exploring data, to be able to just pop coordinates in to get a general idea of terms/functions associated with a brain location. (@adelavega I had wondered about the status of Neurosynth since Tal left academia/cog neuro, and I’m glad to see that there is at least someone tasked with maintaining it…but I realize it’s not as active of a project as it used to be…)

Best,
Michael

I’m happy to say that as a whole it’s very much active (we just got new funding), but most of our bandwith has been focus on the new Neurosynth Compose platform (compose.neurosynth.org) for custom meta-analysis.

We will be soon integrating new AI features to this platform, and have pre-executed meta-analysises like the original Neurosynth, which will then fully replace the legacy platform.

Oohh, cool – good to hear!

In that case, I’ll just put in the request that hopefully the coordinate meta-analysis option will still be available in the new platform… (Even if there are other ways to do it, change is hard. :slight_smile: )

@adelavega Hey I see that the Web interface now works again – thanks for fixing this!

Credit goes to @jdkent. Thank for alerting us!