Brain volumetry: an active contour model-based segmentation followed by SVM-based classification.

Area of Science:

• Medical Imaging
• Neuroscience
• Computer Vision

Background:

• Accurate identification of brain structures in MRI is crucial for volumetric analysis.
• Existing methods may face challenges with segmentation accuracy and efficiency.

Purpose of the Study:

• To present a novel automatic approach for brain structure identification in MRI using active contour models and SVM classifiers.
• To improve the accuracy and efficiency of volumetric measurements through enhanced segmentation.

Main Methods:

• The method employs modified brain images for a novel active contour model, specifically vector field convolution (VFC).
• Simple yet effective features are extracted from segmented images for Support Vector Machine (SVM) classifiers.
• SVM classifiers are trained for gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) identification.

Main Results:

• The VFC approach effectively reduces non-brain regions, preparing images for SVM classification.
• Extracted features proved sufficient and effective for tissue-specific classification.
• Validation using gold standard MRI datasets demonstrated the approach's success compared to existing algorithms.

Conclusions:

• The proposed automatic method offers a robust and accurate solution for brain structure segmentation in MRI.
• This technique holds significant potential for advancing volumetric measurements and neuroimaging research.