Optimizing Energy Landscape Analysis with Genetic Algorithm-Based ROI Selection
Poster No:
1584
Submission Type:
Late-Breaking Abstract Submission
Authors:
Koichiro Mori1, Tomoyuki Hiroyasu1, Satoru Hiwa1
Institutions:
1Doshisha University, Kyotanabe, Kyoto
First Author:
Co-Author(s):
Late Breaking Reviewer(s):
Ruby Kong
Computational Brain Imaging Group, Yong Loo Lin School of Medicine, National University of Singapor
Singapore, Singapore
Computational Brain Imaging Group, Yong Loo Lin School of Medicine, National University of Singapor
Singapore, Singapore
Introduction:
Energy landscape analysis (ELA), which identifies frequent patterns in multidimensional time-series data and models their transitions on energy landscapes using an Ising model, has been applied to neuroimaging data to investigate brain dynamics in a data-driven manner (Ezaki et al., 2017). However, ELA requires pre-determined the region of interest (ROI). This limitation arises because the fitting of the pairwise maximum entropy model (pMEM) is constrained by the number of ROIs required to achieve sufficient accuracy. As a result, ROI selection relies on prior studies or separate analyses, which weakens the data-driven nature of ELA. In this study, we propose ELA/GAopt, which incorporates a genetic algorithm (GA) to search for the optimal combination of ROIs directly from the given data. Furthermore, we evaluate the proposed method using a public dataset on resting-state fMRI and creativity.
Methods:
In our ELA/GAopt, the GA searches for ROIs that exhibit specific dynamic properties, while allowing the selection of any number of regions. The GA iteratively optimizes ROI selection to maximize an objective function combining pMEM fitting accuracy and an index reflecting the targeted brain dynamics. A GA chromosome length corresponds to the total number of brain regions, with each bit indicating whether a region is selected. The pMEM optimization method in the ELA followed Ezaki et al., 2017, and we used a Python implementation of ELA (https://github.com/okumakito/elapy). We implemented pMEM personalization with individualized temperature (Ruffini et al., 2023). With h and J parameters fixed in the model from the group dataset, we optimized temperature parameter β for each individual. Fig. 1 illustrates our framework. To validate our method, a dataset ds002330 was used (Sunavsky & Poppenk, 2019) which includes two 5-min resting-state fMRI scans of 66 healthy adults (26.6 ± 4.3 years). Thirty participants (26.9 ± 4.4 years) were used as the discovery, and 30 remained as an independent testing. In the discovery dataset, GA identified ROIs with specific dynamic properties, such as a negative correlation between the β and the Visual Abbreviated Torrance Test Score for Adults (ATTA) score. These ROIs were then applied to the test dataset to assess whether the same dynamic properties were reproduced. It should be noted that while we used β-ATTA correlation as a criterion for ROI selection, this does not constitute a hypothesis test. The actual hypothesis to be tested using these ROIs should be independently determined by the user and not based on the same criterion used for ROI selection to avoid circular analysis. The maximum number of ROIs was set to 10.
·Fig. 1 Framework of ELA/GAopt.
Results:
The average correlation coefficient between β and ATTA score over 20 GA trials was -0.847, and the fitting accuracy was 0.886 in the discovery dataset. Fig. 2(a) shows the evolution of fitness value in GA optimization, (b) and (c) show the ROIs derived, and (d) presents the local minimum states of the discovery and test dataset.
While the correlation between the model and the creativity score could not be reproduced in the test dataset, the local minimum patterns remained consistent, demonstrating the robustness of identified regions. The optimized ROIs included the regions from DMN, sensorimotor, occipital, and cingulo-opercular networks. The DMN, especially the PCC, plays a key role in creativity. Sensorimotor and occipital regions may support sensory imagery in creative tasks, while the cingulo-opercular network may coordinate cognitive resources for creativity.
While the correlation between the model and the creativity score could not be reproduced in the test dataset, the local minimum patterns remained consistent, demonstrating the robustness of identified regions. The optimized ROIs included the regions from DMN, sensorimotor, occipital, and cingulo-opercular networks. The DMN, especially the PCC, plays a key role in creativity. Sensorimotor and occipital regions may support sensory imagery in creative tasks, while the cingulo-opercular network may coordinate cognitive resources for creativity.
·Fig. 2 Results of the ELA/GAopt on creativity dataset. (a) Evolution of fitness value in GA optimization, (b) and (c) Optimized ROIs, and (d) Local minimum states of the discovery and test dataset.
Conclusions:
Our method identified brain regions associated with creativity and demonstrated the effectiveness of integrating GA with ELA. The consistent local minimum patterns across discovery and test datasets highlight the robustness of this approach. These results emphasize the potential of our ELA/GAopt to capture brain dynamics in a fully data-driven manner, making it a valuable tool for studying brain networks and their dynamics.
Higher Cognitive Functions:
Higher Cognitive Functions Other
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling 2
Methods Development 1
Task-Independent and Resting-State Analysis
Other Methods
Keywords:
Computational Neuroscience
Computing
Data analysis
FUNCTIONAL MRI
Informatics
Machine Learning
Modeling
Other - Creativity;Energy Landscape Analysis
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?
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.
Ruffini, G., Damiani, G., Lozano-Soldevilla, D., Deco, N., Rosas, F. E., Kiani, N. A., Ponce-Alvarez, A., Kringelbach, M. L., Carhart-Harris, R., & Deco, G. (2023). LSD-induced increase of Ising temperature and algorithmic complexity of brain dynamics. PLOS Computational Biology, 19(2), e1010811. https://doi.org/10.1371/journal.pcbi.1010811
Sunavsky, A., & Poppenk, J. (2019). Neuroimaging predictors of creativity in healthy adults. OpenNeuro. [Dataset]. https://doi.org/10.18112/openneuro.ds002330.v1.0.0