Multilevel Thresholding Method Based on Electromagnetism for Accurate Brain MRI Segmentation to Detect White Matter,...

Area of Science:

• Medical Imaging
• Computational Neuroscience
• Biomedical Engineering

Background:

• Accurate brain Magnetic Resonance Imaging (MRI) segmentation is crucial for understanding anatomical structures, identifying abnormalities, and aiding treatment planning, particularly for radiation therapy.
• Existing segmentation methods often face challenges in precisely delineating brain tissues like White Matter (WM), Gray Matter (GM), and Cerebrospinal Fluid (CSF).

Purpose of the Study:

• To develop and validate an advanced and accurate brain MRI segmentation technique.
• To segment brain tissues into WM, GM, and CSF for potential applications in radiation therapy planning.

Main Methods:

• A novel Multilevel Thresholding (MT) method is proposed, integrating an Electromagnetism-Like optimization algorithm with Otsu and Kapur objective functions.
• The method incorporates preprocessing steps including skull stripping and anisotropic diffusion filtering.
• The core segmentation mechanism utilizes attraction-repulsion forces between charged particles to enhance population dynamics.

Main Results:

• The proposed method demonstrated high accuracy in segmenting WM, GM, and CSF from MR brain images.
• Performance evaluation against ground-truth images yielded superior sensitivity, specificity, and segmentation accuracy.
• Comparative analysis against K-means, fuzzy C-means, OTSU MT, PSO, BFA, GA, and FLGMM showed the proposed method's enhanced performance.

Conclusions:

• The Electromagnetism-Like optimized Multilevel Thresholding method offers an advanced and accurate solution for brain MRI segmentation.
• This technique shows significant potential for improving the precision of tissue differentiation in brain MRIs, benefiting clinical applications like radiation therapy planning.