@jasndoc “Total failure” is too sweeping. fMRI helps guide TMS for depression; TMS also has trial evidence for reducing suicidal thoughts. Diffusion MRI informs DBS for OCD, and PET clarifies drug benefits and harms. The translation gap warrants better evidence, not dismissal.
@torwager Maybe the question is whether we have money or it has us: “For the love of money is a root of all kinds of evil.” Or as CS Lewis's Screwtape says: “Prosperity knits a man to the world. He feels that he is finding his place in it, while really it is finding its place in him.”
@PTenigma@gdequeiroz I think polished vs clunky is not the ideal comparison here, rather slides should be a match to your presentation style, and also density of information is important as well. Its why many of us write our own notes for teaching classes rather than using an existing book format.
@Yogi_Spoke@PTenigma large NN may create many near-equivalent good solutions, making it easier for gradient methods to find a compressible model. A smaller network may represent the same solution post hoc but be harder to optimize from scratch. Test case: retrain a pruned network from scratch.
@Yogi_Spoke@PTenigma [late to the party] i think the answer is 'maybe', though it seems the answer for why it still may be necessary to have a larger NN is that the large parameterization can make learning easier, so a small NN may have a similar solution but be way harder to optimize.
@mguindani Yep. Using AI to develop and validate genuinely novel ideas is very different from simply making it possible to produce far more papers (most of which seem to be re-synthesis of existing concepts). I have a doc w/ hundreds of cool project ideas..time to start generating output!
🧠 New in Aperture Neuro:
NeuroFLAME Vaults connects public neuroimaging datasets with private data stored locally, enabling decentralized analyses across sites without sharing sensitive raw data.
https://t.co/fd71LaH0nj
3-day intensive #fMRI course *live online* Sept 9-11, 2026.
#SPM, #ICA, GLM, connectivity, mediation, MRI physics, #DataScience with
@vcalhoun.bsky.social and Kent Kiehl.
We love talking methods & connecting with colleagues! Come join us!
Register here:
https://t.co/F9PB4CUnkc
3-day intensive #fMRI course *live online* Sept 9-11, 2026.
#SPM, #ICA, GLM, connectivity, mediation, MRI physics, #DataScience with
@vcalhoun.bsky.social
and Kent Kiehl.
We love talking methods & connecting with colleagues! Come join us!
Register here:
https://t.co/JJuATJdUIH
Can we map fast neural communication when fMRI scanners sample 20x too slowly?
@Sajad_abv bridges theory & messy clinical data to recover directed brain networks under temporal undersampling in schizophrenia patients and controls.
Correlation says two brain networks are linked. It can't say which one is driving.
New preprint: modeling how slowly fMRI samples the brain recovers the direction — and several times more of what separates schizophrenia patients from controls.
🔗 https://t.co/tWTi868GTP
🚨 Medical AI Research Alert! 🚨
How can we ensure AI agents conducting scientific analyses make 𝘥𝘦𝘧𝘦𝘯𝘴𝘪𝘣𝘭𝘦 𝘤𝘭𝘢𝘪𝘮𝘴 rather than just producing outputs?
@Stanford presents 𝗕𝗿𝗮𝗶𝗻 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵𝗲𝗿, an agentic research harness that embeds methodological judgment directly into neuroimaging data analysis workflows for auditable scientific claims.
By @ZijiaoC, Nicholas Lu, Xinhui Li, @RicardJocelyn, Ce Ju, Huan H. Wang, Christian Kindermann, @mumbrainstats, @StevenDillmann, James Kent, @elneurozorro, @sanmikoyejo, @vdcalhoun, Joshua W. Buckholtz, @HelenJuanZhou , @sbollmann_MRI and @russpoldrack
Now you can watch and listen to the latest Medical AI papers daily on our YouTube and Spotify channels!
YouTube Explainer: https://t.co/hOGzUpjRAK
YouTube Shorts: https://t.co/1klVO9gV6S
Spotify: https://t.co/r8IOqk05Zs
Here's why it's exciting: 👇🧵 1/9
#MedicalAI #Healthcare #NeuroimagingAI #Reproducibility
[1/9]
@Kaju_Nut Some reviewers have a chip on their shoulder/soapbox, so avoiding that is good. But I'm also seeing new (likely AI) reviews that have like 15+ long paragraphs with 10 points requests in each....ridiculous requests equivalent of asking for several new papers.
Join our 22nd event on Sep 25. We're pleased to welcome Dr Souvik Phadikar from @vdcalhoun team at @CenterTrends to introduce his work on "Group Joint ICA (gjICA): Fusing Concurrent EEG and fMRI". Scan QR code to join for free!