A multi-objective optimization approach for brain MRI segmentation using fuzzy entropy clustering and region-based...

Area of Science:

• Medical Imaging
• Computational Biology
• Optimization Algorithms

Background:

• Accurate segmentation of human brain Magnetic Resonance Imaging (MRI) is crucial for neurological studies and clinical diagnosis.
• Existing methods like fuzzy entropy clustering and region-based active contours have limitations that affect segmentation performance.
• There is a need for advanced algorithms that overcome these drawbacks for improved brain MRI analysis.

Purpose of the Study:

• To present a novel multi-objective optimization approach for enhanced brain MRI segmentation.
• To integrate and improve upon fuzzy entropy clustering and active contour methods.
• To achieve superior accuracy and robustness in brain MRI segmentation.

Main Methods:

• A multi-objective particle swarm optimization (MOPSO) approach was employed.
• Two fitness functions, compactness and separation, were derived from kernelized fuzzy entropy clustering with local spatial information and bias correction (KFECSB) and an adaptive energy weight combined with global and local fitting energy active contour (AWGLAC) model.
• A set of non-dominated solutions was optimized simultaneously, with the L2-metric method selecting the best trade-off solution.

Main Results:

• The proposed algorithm demonstrated superior segmentation performance compared to state-of-the-art methods.
• Validation was performed using simulated (BrainWeb) and real (IBSR) MR images.
• Experimental results confirmed enhanced accuracy and robustness in brain MRI segmentation.

Conclusions:

• The developed multi-objective optimization approach offers a significant advancement in brain MRI segmentation.
• The integration of KFECSB and AWGLAC models within MOPSO provides a robust and accurate segmentation technique.
• This method holds promise for improving the analysis of brain structure in medical imaging research.