Related Experiment Video
Updated: Apr 20, 2026
11:50
A Standardized Pipeline for Examining Human Cerebellar Grey Matter Morphometry using Structural Magnetic Resonance Imaging
Published on: February 4, 2022
4.8K
GraSP: geodesic Graph-based Segmentation with Shape Priors for the functional parcellation of the cortex
N Honnorat1, H Eavani1, T D Satterthwaite2
1Center for Biomedical Image Computing and Analytics, Department of Radiology, University of Pennsylvania, Philadelphia, PA 19104, USA.
Neuroimage
|December 3, 2014
Summary
This study introduces a new graph-based brain parcellation method using resting-state functional MRI data. The novel approach improves reproducibility and avoids fragmented brain regions, enhancing functional neuroanatomy mapping.
Area of Science:
- Neuroimaging
- Computational Neuroscience
- Brain Mapping
Background:
- Resting-state functional MRI (rs-fMRI) is crucial for mapping brain function.
- Voxelwise connectivity analysis is often computationally intensive, necessitating dimensionality reduction.
- Existing parcellation methods using anatomic atlases or data-driven approaches have limitations in aligning with functional neuroanatomy and reproducibility.
Purpose of the Study:
- To develop a novel, data-driven graph-based parcellation method for the human cerebral cortex.
- To improve the biological relevance and reproducibility of brain parcellation for connectivity analyses.
- To overcome limitations of existing methods, such as initialization dependency and fragmented parcels.
Main Methods:
- A graph-based parcellation method utilizing a discrete Markov Random Field framework.
- Explicit enforcement of spatial connectedness using shape priors.
- Adaptation of parcel shapes to data via functional geodesic distances, ensuring an initialization-free process.
Main Results:
- The proposed method demonstrates superior reproducibility compared to prevalent parcellation techniques.
- It achieves a similar data fit while avoiding spatially incoherent parcels.
- The approach successfully identifies strong brain developmental effects, which are less apparent with other methods.
Conclusions:
- The novel graph-based parcellation method offers a robust and reproducible way to map brain functional organization.
- It overcomes key limitations of existing methods, providing more biologically grounded and coherent brain parcels.
- This technique enhances the ability to detect subtle but significant neurodevelopmental changes in brain connectivity.
