Precision Confidence Mapping: Finding confidence in individualized functional networks.
Poster No:
1456
Submission Type:
Abstract Submission
Authors:
Julian Ramirez1, Robert Hermosillo2, Julia Moser3, Gracie Grimsrud1, Erendiz Tarakci1, Hannah Pham1, Vanessa Morgan1, Thomas Madison1, Kimberly Weldon1, Oscar Miranda-Dominguez1, Brenden Tervo-Clemmens1, Steven Nelson1, Damien Fair1
Institutions:
1University of Minnesota, Minneapolis, MN, 2Oregon Health & Science University, Portland, OR, 3Masonic Institute for the Developing Brain, University of Minnesota, Minneapolis, MN
First Author:
Co-Author(s):
Julia Moser, PhD
Masonic Institute for the Developing Brain, University of Minnesota
Minneapolis, MN
Masonic Institute for the Developing Brain, University of Minnesota
Minneapolis, MN
Introduction:
There has been a shift in resting-state, rs-fMRI from relying on group-averaged functional network topographies toward subject-specific configurations derived from densely sampled scans. Although existing approaches can identify networks within individual participants (e.g., Hermosillo et al. 2024; Laumann et al. 2015), they yield categorical network assignments that do not estimate uncertainty and measurement error. In scenarios demanding precise, individualized network localization-such as neuromodulation-categorical-only labels become problematic for subjects with limited data. This underscores the need for confidence measures, as smaller datasets with short acquisitions (common in clinical applications) yield less stable network assignments that compromise reliability.
Methods:
We introduce Precision Confidence Mapping (PCM), an extension of network-detection methods that provides individualized functional network topographies and their confidence maps. PCM leverages the temporal nature of rs-fMRI data by bootstrapping volumes (e.g., n=100 iterations) into new, shuffled time series of a user-defined length. A community detection algorithm is applied to each bootstrap iteration, yielding a probability of network assignments for each grayordinate. Assignments are aggregated across permutations to produce "confidence maps" that estimate network stability across iterations. This framework enables analyses of individual-specific functional networks and facilitates analyses of network stability, data requirements, and assignment probabilities. Further, confidence maps can be thresholded, enabling researchers to identify regions with differing degrees of assignment certainty.
Results:
We validated PCM on four adults, each contributing 140 minutes of low-motion (FD<0.2mm) multi-band, multi-echo rsfMRI data. Each dataset was divided into two 70-minute halves: one served as a "ground truth" validation, while the other ("exploratory") was split into bins ranging from 2 to 70 minutes to examine the effect of rs-fMRI duration on confidence maps. We examined both dice coefficients and positive predictive value (PPV) between the exploratory bins and ground truth to assess spatial overlap versus assignment accuracy. With short durations, PCM reveals network assignments with moderate to high uncertainty, which decreases as dataset length increases. Unthresholded confidence maps showed dice coefficients increased from 0.26±0.13 at 2 minutes to 0.74±0.08 at 70 minutes. Applying thresholds to confidence maps (Figure 1A & D) revealed that lower thresholds (10–30%) yielded higher dice scores (e.g., 0.48±0.08 at 2 minutes and 0.75±0.07 at 70 minutes). However, as thresholds increased this dice score recovery declined (Figure 1C). To capture how accurately each thresholded exploratory map aligned with ground truth network labels, we calculated PPV (Figure 1E). For a 10% threshold at 2 minutes, the mean PPV was 0.52±0.13, rising to 0.77±0.10 at 70 minutes. However, a 90% threshold at 2 minutes yielded a higher PPV of 0.85±0.12, more closely matching the 70-minute bin (0.78±0.11). This underscores how thresholding retains accurate vertices in short, variable datasets, while minimally affecting stable assignments at longer durations. These results suggest that while lower thresholds improve overall spatial overlap (dice), higher thresholds increase accuracy (PPV) of identified regions, which could be crucial for targeted interventions.
·Precision confidence maps illustrating how data length and thresholding improve assignment stability and accuracy.
Conclusions:
These findings illustrate how PCM refines individualized network identification by quantifying stability for both data quantity and confidence thresholds. Unlike purely binary assignment methods, PCM offers a tunable measure of certainty that can inform precision interventions (e.g., Deep Brain Stimulation, Transcranial Magnetic Stimulation) and refine minimum data requirements for reliable network mapping. As PCM evolves, it promises to integrate precise neuroscientific insights into personalized therapies.
Brain Stimulation:
Non-Invasive Stimulation Methods Other 2
Modeling and Analysis Methods:
fMRI Connectivity and Network Modeling 1
Segmentation and Parcellation
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping
Novel Imaging Acquisition Methods:
BOLD fMRI
Keywords:
Computational Neuroscience
Cortex
Data analysis
Design and Analysis
FUNCTIONAL MRI
Informatics
Segmentation
Statistical Methods
Systems
Transcranial Magnetic Stimulation (TMS)
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Are you Internal Review Board (IRB) certified? Please note: Failure to have IRB, if applicable will lead to automatic rejection of abstract.
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
Hermosillo, R. J. M., Moore, L. A., Feczko, E., Miranda-Domínguez, Ó., Pines, A., Dworetsky, A., Conan, G., Mooney, M. A., Randolph, A., Graham, A., Adeyemo, B., Earl, E., Perrone, A., Morales Carrasco, C., Uriarte-Lopez, J., Snider, K., Doyle, O., Cordova, M., Koirala, S., … Fair, D. A. (2024). A precision functional atlas of personalized network topography and probabilities. Nature Neuroscience, 27, 1000–1013.