Brain Tumor Segmentation Based on Hybrid Clustering and Morphological Operations

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Biomedical Engineering

Background:

• Brain tumor segmentation from MRI is challenging due to complex structures, blurred boundaries, and noise.
• Existing methods often suffer from noise sensitivity and parameter instability.

Purpose of the Study:

• To propose an effective hybrid clustering algorithm for robust brain tumor segmentation.
• To improve the accuracy, stability, and noise resilience of brain tumor segmentation from MRI data.

Main Methods:

• Adaptive Wiener filtering for denoising MRI data.
• Hybrid clustering combining K-means++ and Gaussian kernel-based fuzzy C-means for image segmentation.
• Morphological operations and median filtering for preprocessing and postprocessing to refine tumor boundaries.

Main Results:

• The proposed algorithm demonstrated reduced sensitivity to noise and improved segmentation stability.
• Postprocessing steps effectively refined tumor representations.
• Comparative analysis showed superior performance in accuracy, sensitivity, specificity, and recall over existing methods.

Conclusions:

• The hybrid clustering algorithm offers a robust and accurate solution for brain tumor segmentation in MRI.
• The integration of adaptive filtering and morphological operations enhances the reliability of the segmentation process.