Constructing Morphometric INverse Divergence (MIND) Networks for the Developing Human Brain
Poster No:
951
Submission Type:
Abstract Submission
Authors:
Ashleigh Davies1, Isaac Sebenius2, Vyacheslav Karolis3, Tomoki Arichi3, Sarah Morgan1,4
Institutions:
1Department of Biomedical Computing, BMEIS, King's College London, London, 2Department of Psychiatry, Cambridge University, Cambridge, 3Early Life Imaging Research Department, BMEIS, King's College London, London, 4Department of Computer Science and Technology, Cambridge University, Cambridge
First Author:
Co-Author(s):
Sarah Morgan
Department of Biomedical Computing, BMEIS, King's College London|Department of Computer Science and Technology, Cambridge University
London|Cambridge
Department of Biomedical Computing, BMEIS, King's College London|Department of Computer Science and Technology, Cambridge University
London|Cambridge
Introduction:
Structural similarity has been identified as a key indicator of brain development– with Morphometric Similarity Networks (MSNs) demonstrated to accurately predict neurodevelopmental outcomes from neonatal brain images (Fenchel, 2022). We recently proposed a new technique to estimate structural similarity: Morphometric INverse Divergence (MIND), which determines structural similarity from multivariate distributions of vertex-level features on the cortical surface (Sebenius, 2023). MIND networks have proven to be a highly effective tool for representing cytoarchitecture and connectivity in the adult brain, as well as for the prediction of brain development from 8-21 years, outperforming MSNs for age prediction and correlating more strongly with gold standard tract tracing in the macaque.
Methods:
Here, we extend MIND to study early human brain development, by constructing MIND networks for a subset of N=179 neonates (mean age of 40.0 weeks; std = 2.53) from the developing Human Connectome Project (dHCP) (Edwards, 2022). T1- and T2-weighted Magnetic Resonance Images (MRI) were pre-processed, and cortical meshes generated using neonatal cortical surface reconstruction (Makropoulos, 2018) and multimodal surface matching (Bozek 2018). Cortical regions were defined using the DrawEM neonatal atlas (Adamson, 2020). Cortical surface feature maps were derived for cortical thickness, mean curvature, sulcal depth and a proxy measure of myelination (T1/T2 ratio). We derived multivariate distributions from these four features, and analysed them at the vertex level using k-nearest-neighbour estimation for the inverse KL Divergence between multivariate distributions for each possible pair of regions (Perez-Cruz, 2008). Matrices containing pairwise regional similarity values were derived, producing a multimodal MIND network for each neonate.
Patterns of regional MIND values were analysed for inter-subject consistency by calculating the pairwise Pearson correlation between flattened upper triangles of subject-level MIND networks. Global average similarity values were also calculated for the MIND network and four single modality feature networks for each neonate, and compared to their post menstrual age (PMA) at scan to observe changes in inter-regional similarity relating to early morphometric brain development.
Patterns of regional MIND values were analysed for inter-subject consistency by calculating the pairwise Pearson correlation between flattened upper triangles of subject-level MIND networks. Global average similarity values were also calculated for the MIND network and four single modality feature networks for each neonate, and compared to their post menstrual age (PMA) at scan to observe changes in inter-regional similarity relating to early morphometric brain development.
Results:
Multimodal MIND networks were successfully constructed for all 179 participants (cohort mean MIND = 0.09; std = 0.004). Figure 1a shows the MIND network of a representative neonate (n=1), arranged according to 8 major cortical areas. High regional similarity was found amongst regions within the frontal, parietal and occipital areas, with inter-subject consistency high across all regions (mean r = 0.9049; std = 0.0358) (Figure 1b). Average MIND similarity was found to increase significantly with post menstrual age (r = 0.62) (Figure 2). MIND networks constructed from individual features showed a similar trend of increasing similarity with post menstrual age for cortical thickness (r = 0.66), mean curvature (r = 0.79) and sulcal depth (r = 0.82). Single modality networks constructed from T1/T2 ratio only had a trend towards decreased similarity with age (r = -0.38), suggesting that MIND networks are sensitive to timing of different developmental processes.
Conclusions:
In sum, the construction of neonatal MIND networks opens the door for multidimensional, multi-feature analysis of early life brain development, building on existing understanding of the importance of structural similarity. The unique ability of MIND to prioritise cortical features that best differentiate cortical regions has promising implications for identifying and predicting altered development trajectories. This initial analysis of healthy neonates across a range of gestational and premenstrual ages serves as a baseline from which to investigate patterns of brain development relating to neurodevelopmental disorders.
Lifespan Development:
Early life, Adolescence, Aging 1
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping 2
Novel Imaging Acquisition Methods:
Anatomical MRI
Keywords:
Development
MRI
Multivariate
STRUCTURAL MRI
Other - Structural Similarity
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?
Provide references using APA citation style.
2. Bozek, J. (2018). Construction of a neonatal cortical surface atlas using multimodal surface matching in the developing human connectome project. NeuroImage. 179, 11-29.
3. Edwards, A.D. (2022). The developing human connectome project neonatal data release. Frontiers. 16.
4. Fenchel, D. (2022). Neonatal multi-modal cortical profiles predict 18-month developmental outcomes. Developmental Cognitive Neuroscience. 54, 101-103.
5. Makropoulos, A. (2018). The developing human connectome project: a minimal processing pipeline for neonatal cortical surface reconstruction. NeuroImage. 173, 88-112.
6. Perez-Cruz, F. (2008). Kullback-Leibler divergence estimation of continuous distributions. IEEE International Symposium on Information Theory. 1666-1670.
7. Sebenius, I. (2023). Robust estimation of cortical similarity networks from brain MRI. Nature Neuroscience, 26, 1461-1471.