An Advanced MRI Multi-Modalities Segmentation Methodology Dedicated to Multiple Sclerosis Lesions Exploration and...

Area of Science:

• Biomedical imaging
• Neurology
• Medical image analysis

Background:

• Multiple sclerosis (MS) is a common neurological disease affecting young individuals.
• Accurate segmentation of MS lesions is crucial for diagnosis and management.
• Existing methods face challenges in precise lesion identification and characterization.

Purpose of the Study:

• To develop an automatic biomedical aided tool for volumetric segmentation of multiple sclerosis lesions.
• To improve the accuracy and efficiency of MS lesion detection and characterization in MRI scans.
• To provide a tool that assists clinicians in MS diagnosis and longitudinal monitoring.

Main Methods:

• Preliminary cerebral zones segmentation using a novel Gaussian mixture model.
• Lesion segmentation involving lesion map estimation, threshold constraints, and a new lesion expansion algorithm.
• Validation on four diverse clinical MRI databases with varying lesion loads and noise levels.

Main Results:

• Excellent cerebral segmentation with Dice averages near 0.8 and sensitivity/specificity > 0.9.
• Lesion segmentation metrics (Dice, sensitivity, specificity) averaged >= 0.8 across databases.
• Accurate identification of lesion cores and surrounding vasogenic edema, validated by experts.

Conclusions:

• The proposed methodology offers a robust and accurate tool for MS lesion segmentation.
• The computer-aided diagnosis tool can significantly aid clinicians in MS diagnosis and patient monitoring.
• The approach shows potential for application in other MRI-based neurological diseases like glioblastoma and Alzheimer's disease.