Related Experiment Video
Updated: May 10, 2026
12:30
A Comprehensive Protocol for Manual Segmentation of the Medial Temporal Lobe Structures
Published on: July 2, 2014
Improved inference in Bayesian segmentation using Monte Carlo sampling: application to hippocampal subfield volumetry
Juan Eugenio Iglesias1, Mert Rory Sabuncu, Koen Van Leemput
1Martinos Center for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School, USA. iglesias@nmr.mgh.harvard.edu
Medical Image Analysis
|June 19, 2013
Summary
This study introduces a new Bayesian approach for medical image segmentation, improving accuracy by accounting for parameter uncertainty. This enhances Alzheimer's disease classification using brain MRI scans.
Area of Science:
- Medical Image Analysis
- Computational Anatomy
- Bayesian Statistics
Background:
- Medical image segmentation often uses Bayesian models with fixed parameters.
- Estimating parameters and fixing them ignores their inherent uncertainty.
- This limitation can affect the accuracy of segmentation results.
Purpose of the Study:
- To improve Bayesian segmentation methods by incorporating parameter uncertainty.
- To enhance the inference process in medical image analysis.
- To refine segmentation of brain structures for disease classification.
Main Methods:
- Incorporating parameter uncertainty into Bayesian segmentation models.
- Approximating marginalization over model parameters using Markov chain Monte Carlo (MCMC) techniques.
- Applying the method to segment hippocampal subfields in brain MRI scans.
Main Results:
- Demonstrated significant improvement in an Alzheimer's disease classification task.
- Showcased enhanced accuracy in segmenting hippocampal subfields.
- Enabled computation of informative error bars for volume estimates of segmented structures.
Conclusions:
- Accounting for parameter uncertainty in Bayesian segmentation improves accuracy.
- The proposed MCMC-based approach offers a computationally efficient solution.
- This method has potential applications in neuroimaging and disease diagnosis.
