Related Experiment Video
Updated: May 31, 2026
11:03
High-resolution In Vivo Manual Segmentation Protocol for Human Hippocampal Subfields Using 3T Magnetic Resonance Imaging
Published on: November 10, 2015
Appearance-based modeling for segmentation of hippocampus and amygdala using multi-contrast MR imaging
Shiyan Hu1, Pierrick Coupé, Jens C Pruessner
1McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University, Montréal, Québec, Canada H3A 2B4. shiyanhu99@yahoo.com
Neuroimage
|July 12, 2011
Summary
This study introduces an automatic MRI segmentation method using active appearance modeling (AAM) and level sets. It improves segmentation of the hippocampus and amygdala by incorporating multi-contrast images and non-linear alignment.
Area of Science:
- Medical imaging analysis
- Computational anatomy
- Machine learning in medicine
Background:
- Accurate segmentation of brain structures like the hippocampus and amygdala is crucial for neurological studies.
- Traditional segmentation methods struggle with structures exhibiting low contrast in MRI scans.
Purpose of the Study:
- To develop and evaluate an automatic model-based segmentation scheme for human hippocampi and amygdalae.
- To enhance segmentation performance by integrating multi-contrast MRI data and optimizing active appearance modeling (AAM).
Main Methods:
- A novel automatic segmentation approach combining level set shape modeling and active appearance modeling (AAM).
- Incorporation of multi-contrast MRI (T1, T2, proton density) with optimized contrast weighting within AAM.
- Comparison of linear and non-linear alignment strategies for training data, focusing on local volumes of interest.
Main Results:
- Non-linear alignment of training data significantly improved segmentation accuracy.
- Multimodal segmentation using combined contrasts outperformed single-contrast segmentation.
- The method achieved high accuracy for hippocampus (mean Dice κ=0.87, ICC=0.946) and amygdala (mean Dice κ=0.81, ICC=0.924) segmentation in cross-validation.
Conclusions:
- The proposed model-based segmentation scheme effectively segments hippocampi and amygdalae, even with weak intensity contrast.
- Optimized multi-contrast AAM with non-linear alignment represents a robust approach for accurate brain structure segmentation.
