Related Experiment Video
Updated: Apr 10, 2026
Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
Published on: January 7, 2019
Deep infant brain segmentation from multi-contrast MRI
Malte Hoffmann1,2,3, Lilla Zöllei1,2,3, Adrian V Dalca1,2,3,4
1Athinoula A. Martinos Center for Biomedical Imaging.
Insights
BabySeg is a new deep learning framework for segmenting infant and child brain MRIs. It overcomes challenges in pediatric neuroimaging, offering accurate results across diverse scan types and ages.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Pediatric Radiology
Background:
- Accurate segmentation of pediatric brain MRI is crucial for studying brain development.
- Challenges include developmental changes, imaging constraints, motion artifacts, and diverse protocols.
- Existing segmentation models are often specialized and lack robustness for clinical variability.
Purpose of the Study:
- To develop a versatile deep learning framework for robust brain MRI segmentation in infants and young children.
- To address the fragmentation of existing specialized segmentation methods.
- To improve segmentation accuracy and efficiency across diverse pediatric neuroimaging data.
Main Methods:
- Developed BabySeg, a deep learning framework utilizing domain randomization to synthesize varied training images.
- Implemented a feature pooling mechanism allowing models to integrate information from multiple input scans.
- Trained and validated the framework on diverse pediatric MRI datasets with varying protocols and age groups.
Main Results:
- BabySeg achieved state-of-the-art performance in segmenting pediatric brain MRIs.
- The single model demonstrated comparable or superior accuracy to existing methods across different age cohorts and input configurations.
- The framework processed images significantly faster than many existing tools.
Conclusions:
- BabySeg offers a unified and robust solution for pediatric brain MRI segmentation, overcoming limitations of specialized models.
- The framework's adaptability to diverse imaging protocols and its high accuracy make it suitable for clinical and research applications.
- This approach advances the analysis of early human brain development through improved neuroimaging segmentation.
Abstract:
Segmentation of magnetic resonance images (MRI) facilitates analysis of human brain development by delineating anatomical structures. However, in infants and young children, accurate segmentation is challenging due to development and imaging constraints. Pediatric brain MRI is notoriously difficult to acquire, with inconsistent availability of imaging modalities, substantial non-head anatomy in the field of view, and frequent motion artifacts. This has led to specialized segmentation models that are often limited to specific image types or narrow age groups, or that are fragile for more variable images such as those acquired clinically. We address this method fragmentation with BabySeg, a deep learning brain segmentation framework for infants and young children that supports diverse MRI protocols, including repeat scans and image types unavailable during training. Our approach builds on recent domain randomization techniques, which synthesize training images far beyond realistic bounds to promote dataset shift invariance. We also describe a mechanism that enables models to flexibly pool and interact features from any number of input scans. We demonstrate state-of-the-art performance that matches or exceeds the accuracy of several existing methods for various age cohorts and input configurations using a single model, in a fraction of the runtime required by many existing tools.
