Dan Handwerker

We’re hiring! Pls share widely:

In conjunction w/ the @bowers-wbhi.bsky.social, my team is recruiting a full-time clinical research coordinator to assist with research efforts at UC Santa Barbara. 1-2 yrs experience in MRI data collection preferred.

Submit apps to: recruit.ap.ucsb.edu/JPF02974

Maybe @cwindfmri.bsky.social @tmsfmri.bsky.social can suggest some most current TMS safety literature.

Moving beyond processing- and analysis-related variation in resting-state functional brain imaging - Nature Human Behaviour Functional connectivity estimates vary significantly across different functional magnetic resonance imaging preprocessing pipelines. Due to these variations, using seemingly similar minimal preprocess...

The flatness to 1000 parcels does surprise me because it's slightly in conflict with doi.org/10.1038/s415... where I think their main effect is caused by instability of group anat alignment increasing with smaller ROIs. 3/3

For ROIs, depending on output measure, I suspect there's a balance between SNR from averaging, whether one is averaging distinct areas together, & when ROIs are so small that they don't reliably represent the same areas across the population. 2/3

Even though the data may support it, I think few people use flip angle <30, but I do use well lower than Ernst angle. A main take-away is that TNSR is flat across a wide range of SNRs, but it does drop with very low SNR and drop point will depend on other factors like voxel size & echo time. 1/3

Just to be clear, this paper really is great & important work. If what I'm saying is even a critique, it's pushing at potential ways to build on top of this work. 4/4

For example for an ROI based RSFC metric, factors I'd expect to matter include levels of (1) Non-neural physiological noise, (2) Signal dropout, (3) Spatial distortion / mis-alignment. We don't need to solve for all, but would need to quantify application specific CNR improvements. 3/4

The changes (acquisition or processing) that I would expect to matter are ones that affect a contrast-of-interest / noise rather than a general SNR improvement. 2/4

I think one part that makes this hard is that may ways to improve SNR increase both scanner and physiological noise (i.e. flip angle: doi.org/10.1016/j.ne... ). Also, since I think this work is mostly (all?) using ROIs, then some SNR changes in voxel size are reduced through voxel averaging. 1/4

"Precision" is essential for all cool new methods. 40 minutes is a lot of data so maybe this should be "Super Precision". This is based on task data so, even if regressed out, it should mention there's "Activity". I think "Super Precision Activity Mapping" might be a better name. 😀

Happy to talk more with @bttyeo.bsky.social or others in this thread, if there's interest. 4/4

It wouldn't matter if data quality was improved through changes to acquisition, processing/denoising, task design, or output measures-of-interest. As long as the improvement could be quantified in relationship to the data used for the base model, it could fit into this model. 3/4

The existing method is testing 1 parameter of intra-participant improvement: scan length. If their model has a baseline data quality expectation (defined by a CNR or test-retest stability measure), then others could pilot scans/analyses and plug their relative CNR change into the same model. 2/4

I think extending this tool for data quality is non-trivial, but doesn't require modeling every type of data quality improvement. At the core, this method is saying, if you improve intra-participant data quality, you can get similar quality results with fewer total participants. 1/4

Just you.

thomasyeolab.github.io/OptimalScanT... is an awesome tool and a service to the neuroimaging community, but I am concerned if our necessary data size calculations are based on only two study's choices about acquisition priorities and preprocessing pipelines. 2/2

Great & important work. That said, one area that is still overlooked is data quality. This is harder to model because contrast-to-noise can be application specific, but it's critical when designing large studies. Better data means similar results with smaller N or shorter scans. 1/2

Having severe FOMO for missing #OHBM2025, but please reach out if you are interested in postdoc opportunities in our lab! Enjoy the science and sunshine!!

Single slice EPI to visualize brain pulsations

practicalfmri.blogspot.com/2025/06/sing...

Interested in practical tips for assessing collinearity and why common fixes might not help? Come to the #OHBM2025 oral session Friday at 11:30AM (P2)! I’ll share simulation results and @demidenko.bsky.social will present real data. Also, check out my poster Thursday (1051)!

#OHBM2025 multi-echo fMRI users meet-up at 7:30PM on Thursday, June 26th very close to the convention center. DM me for location information.

Also great #OHBM2025 talks by Leigh Sepeta on imaging in pediatric epilepsy populations, with some focus on memory systems and deficits, David Vaughn on individualizing clinical fMRI protocols, and Shalini Narayana on uses of clinical TMS for language mapping in epilepsy.

A slide from a presentation by Chris Tailby. It shows their metric of language lateralization in a normative population that can be used to flag non-normative lateralization patterns in patient who can then be flagged for manual review.

Great session at #OHBM2025 on Clinical applications of brain mapping across the lifespan. Chris Tailby's summary of www.epilepsyproject.org.au is a great example of type of large-scale work that's necessary to make fMRI useful in normative clinical practice.

OHBM 2025 multi-echo meetup

As in past years, I am helping plan a multi-echo fMRI users meet-up at #OHBM2025. There might also be some multi-echo fMRI presence at the preceding Brainhack. If you are interested in joining either, please fill out this form.

GitHub - ME-ICA/ohbm-2025-multiecho: Multi-echo related content at OHBM 2025 Multi-echo related content at OHBM 2025. Contribute to ME-ICA/ohbm-2025-multiecho development by creating an account on GitHub.

Here's a list of content related to multi-echo fMRI at #OHBM2025 Let me know if there's anything to add. github.com/ME-ICA/ohbm-...

Home What can we do with functional MRI data? For many years, fMRI has been used to discover population-level patterns of brain function, organization and connectivity, and to understand differences in t...

Looking forward to #OHBM2025 in Brisbane next week! My lab is recruiting a postdoc & neuroimaging analyst/developer to support our NIH funded work in fMRI-based Alzheimer's biomarker development (NIA R01AG083919). Email or DM if you want to meet up in Brisbane! www.statmindlab.com/join-us

I've seen the Open Science Room at @ohbmofficial.bsky.social evolve over years. The OSR improves the meeting & is an essential element for building community, particularly for trainees. It also causes logistical problems. The push/pull to improve positives while addressing problems should continue.

OHBM 2025 multi-echo meetup

As in past years, I am helping plan a multi-echo fMRI users meet-up at #OHBM2025. There might also be some multi-echo fMRI presence at the preceding Brainhack. If you are interested in joining either, please fill out this form.

Bug found and fixed in tedana v.25.0.0. If you are using that version, upgrade to 25.0.1. groups.google.com/g/tedana-new...

Check out the Tedana multi-echo denoising pipeline develop for EPTI, by Katie Lamar and Dr. Tom Liu at UCSD, presented in multi-echo fMRI users meeting!