Systematic comparison of resting vs. task brain parcellation for precision functional mapping
Poster No:
1635
Submission Type:
Abstract Submission
Authors:
Da Zhi1,2, Caroline Nettekoven3, Ladan Shahshahani4, Ana Luísa Pinho3, Jörn Diedrichsen3, Tian Ge1,2
Institutions:
1Massachusetts General Hospital, Boston, MA, 2Harvard Medical School, Boston, MA, 3Western University, London, Ontario, 4Brown University, Providence, RI
First Author:
Co-Author(s):
Introduction:
Current efforts to develop comprehensive brain parcellations are limited by the availability of large, homogeneous brain imaging datasets. While several atlases have been derived from resting-state fMRI data to capture intrinsic brain organization (Yeo, 2011), a comprehensive task-based parcellation remains missing. Importantly, the functional topography of the brain varies across different states (Salehi, 2020), underscoring the need for a systematic investigation of brain organization during both rest and task activity for precise functional localization. Here, we present novel cortical parcellations derived from both rest and task fMRI, using a hierarchical Bayesian parcellation framework (Zhi, 2024). The task parcellation was generated from a diverse set of fMRI datasets, spanning a wide range of functional task domains. We demonstrate that, compared to the resting-state parcellation, the task parcellation (a) preserves the functional organization observed in the resting-state atlas at the population level, and (b) outperforms existing parcellations in localizing cognitive tasks at the individual level.
Methods:
The task parcellation was trained on six fMRI datasets: Highres MDTB (unpublished), Nakai & Nishimoto (Nakai, 2020), IBC (Pinho, 2018), WMFS (Shahshahani, 2024), Multi-Demand (Assem, 2024), and Somatotopic (Saadon-Grosman, 2022). The rest parcellation was trained solely on the HCP dataset. We used the MDTB (N=24) dataset (King, 2019) throughout as an independent test set, given its broad coverage of cognitive tasks.
All parcellations were generated using a hierarchical Bayesian parcellation framework (Zhi, 2024). The task group parcellation (Fig.1d) was trained by integrating the six training datasets, while the rest group parcellation (Fig.1c) was derived from the HCP dataset. At the individual level, we inferred parcellations for each test subject by adaptively applying a new spatial model (m-RBM, Fig.1a) within the framework. To ensure a fair comparison, the number of parcels for all generated parcellations was set to K=17.
The evaluation metrics used in this work included the Dice coefficient, DCBC (Zhi, 2024), and task inhomogeneity. The Dice coefficient was used to assess the network correspondence between the proposed parcellations with an existing atlas (Yeo, 2011). The DCBC method measures how well a parcellation aligns with the functional boundaries in the test set, while task inhomogeneity quantifies the consistency with the activation of a region during tasks.
All parcellations were generated using a hierarchical Bayesian parcellation framework (Zhi, 2024). The task group parcellation (Fig.1d) was trained by integrating the six training datasets, while the rest group parcellation (Fig.1c) was derived from the HCP dataset. At the individual level, we inferred parcellations for each test subject by adaptively applying a new spatial model (m-RBM, Fig.1a) within the framework. To ensure a fair comparison, the number of parcels for all generated parcellations was set to K=17.
The evaluation metrics used in this work included the Dice coefficient, DCBC (Zhi, 2024), and task inhomogeneity. The Dice coefficient was used to assess the network correspondence between the proposed parcellations with an existing atlas (Yeo, 2011). The DCBC method measures how well a parcellation aligns with the functional boundaries in the test set, while task inhomogeneity quantifies the consistency with the activation of a region during tasks.
Results:
We first assessed the similarity between the proposed parcellations and the Yeo 17 (Fig.1b) parcellation using the Dice coefficient. The results showed that the functional regions in proposed parcellations largely corresponded to the target networks (Fig.1e,f). This suggests that brain functional organization during task activation shares considerable overlap with that of the rest. Furthermore, the evaluation results suggested that both proposed parcellations outperformed the the Yeo 17 atlas (Fig.2a,b).
Next, we estimated two sets of individual parcellations from either rest or task group atlas (Fig.2c) using the m-RBM model on MDTB dataset. We then evaluated their performance using data from the same subjects. The results suggested that the task-based parcellation outperformed its resting-state counterpart in localizing functional regions at the individual level (Fig.2d,e).
Next, we estimated two sets of individual parcellations from either rest or task group atlas (Fig.2c) using the m-RBM model on MDTB dataset. We then evaluated their performance using data from the same subjects. The results suggested that the task-based parcellation outperformed its resting-state counterpart in localizing functional regions at the individual level (Fig.2d,e).
Conclusions:
In this work, we presented a task-based parcellation learned from multiple fMRI datasets using a hierarchical Bayesian framework. Compared to resting-state parcellation, the proposed task-based parcellation: (a) shows similar but not identical functional organization and performs better in localizing task-related functional regions; (b) provides more precise individual functional localizers when localizing cognitive tasks. We believe this systematic comparison provides practical insights into how the choice of data type can influence the precision of individual functional localization.
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI) 2
Bayesian Modeling
fMRI Connectivity and Network Modeling
Segmentation and Parcellation 1
Neuroinformatics and Data Sharing:
Brain Atlases
Keywords:
Atlasing
Computational Neuroscience
Cortex
Data analysis
Machine Learning
Modeling
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.
Cole, M. W. (2014). Intrinsic and task-evoked network architectures of the human brain. Neuron, 83(1), 238–251.
King, M. (2019). Functional boundaries in the human cerebellum revealed by a multi-domain task battery. Nature Neuroscience, 22(8), 1371–1378.
Nakai, T. (2020). Quantitative models reveal the organization of diverse cognitive functions in the brain. Nature Communications, 11(1), 1–12.
Pinho, A. L. (2018). Individual Brain Charting, a high-resolution fMRI dataset for cognitive mapping. Scientific Data, 5(1), 1–15.
Saadon-Grosman, N. (2022). A third somatomotor representation in the human cerebellum. Journal of Neurophysiology, 128(4), 1051–1073.
Salehi, M. (2020). There is no single functional atlas even for a single individual: Functional parcel definitions change with task. NeuroImage, 208, 116366.
Shahshahani, L. (2024). Selective recruitment of the cerebellum evidenced by task-dependent gating of inputs. ELife, 13, RP96386.
Yeo, B. T. T. (2011). The organization of the human cerebral cortex estimated by intrinsic functional connectivity. Journal of Neurophysiology.
Zhi, D. (2024). A hierarchical Bayesian brain parcellation framework for fusion of functional imaging datasets. Imaging Neuroscience.