Related Experiment Videos
Automated segmentation of multispectral brain MR images
Anders H Andersen1, Zhiming Zhang, Malcolm J Avison
1Department of Anatomy and Neurobiology, University of Kentucky Medical Center, 800 Rose Street, Lexington, KY 40536-0098, USA. anders@mri.uky.edu
Journal of Neuroscience Methods
|January 22, 2003
Summary
This study introduces automated brain segmentation for measuring gray matter, white matter, and cerebrospinal fluid using multispectral MRI. The method accurately quantifies tissue volumes in diverse brain imaging data.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Computational Neuroscience
Background:
- Accurate segmentation of brain tissues is crucial for quantitative analysis in neuroscience and clinical research.
- Existing methods often struggle with multispectral MRI data variability and artifacts like radio frequency field inhomogeneity.
- Automated and robust techniques are needed for reliable in vivo brain composition measurement.
Purpose of the Study:
- To develop and validate a comprehensive automated approach for in vivo brain segmentation and quantitative tissue volume measurement.
- To address challenges posed by heterogeneous multispectral MRI data, including varying contrast, intensity weighting, spatial resolution, and orientation.
- To accurately partition intracranial volume into gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) spaces.
Main Methods:
- Utilized statistical pattern recognition with a finite mixture model for brain tissue classification.
- Implemented a masking algorithm for efficient extraction of brain volume from extrameningeal tissues.
- Employed a recursive method to correct for radio frequency (RF) field inhomogeneity, preserving local tissue contrast.
Main Results:
- Demonstrated robust performance in segmenting and quantifying brain tissue volumes across diverse datasets.
- Successfully handled multispectral T1-, proton density-, and T2-weighted MRI data.
- Validated the technique on data from both non-human primates (young and aged) and human subjects.
Conclusions:
- The proposed method offers a robust and comprehensive solution for automated in vivo brain segmentation and quantitative tissue volume measurement.
- The technique effectively handles heterogeneous multispectral MRI data, improving reliability and accuracy.
- This approach has significant potential for applications in neuroscience research and clinical diagnostics requiring precise brain composition analysis.