Related Experiment Video
Updated: Dec 27, 2025
09:57
How to Measure Cortical Folding from MR Images: a Step-by-Step Tutorial to Compute Local Gyrification Index
Published on: January 2, 2012
28.4K
Harmonization of Infant Cortical Thickness Using Surface-to-Surface Cycle-Consistent Adversarial Networks
Fenqiang Zhao1,2, Zhengwang Wu2, Li Wang2
1Key Laboratory of Biomedical Engineering of Ministry of Education, Zhejiang University, Hangzhou, China.
Medical Image Computing and Computer-Assisted Intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention
|March 5, 2020
Summary
This study introduces a novel method using spherical CycleGAN to harmonize infant brain MRI data, effectively removing scanner-specific variations while preserving individual brain development details.
Area of Science:
- Neuroimaging
- Developmental Neuroscience
- Medical Image Analysis
Background:
- Multi-site infant neuroimaging datasets are crucial for studying early brain development.
- Differences in MRI scanners introduce non-biological variance, complicating joint data analysis.
- Harmonizing cortical thickness maps across scanners is essential for accurate research.
Purpose of the Study:
- To develop a method for harmonizing cortical thickness maps from different MRI scanners.
- To address the challenge of non-biological variance in multi-site infant neuroimaging.
- To preserve individual differences in brain development during harmonization.
Main Methods:
- Proposed a surface-to-surface CycleGAN (S2SGAN) combining spherical U-Net and CycleGAN.
- Modeled harmonization as a surface-to-surface translation task.
- Utilized cycle consistency and correlation coefficient loss for accurate harmonization.
Main Results:
- The S2SGAN method effectively harmonizes cortical thickness maps across different scanners.
- The method successfully removes unwanted scanner effects.
- Individual differences in cortical thickness are preserved post-harmonization.
Conclusions:
- The proposed S2SGAN method demonstrates superior performance in harmonizing infant brain MRI data.
- This approach enhances the reliability of multi-site infant neuroimaging studies.
- The method effectively balances scanner harmonization with the preservation of biological variance.
