Conditional spatial biased intuitionistic clustering technique for brain MRI image segmentation.

Area of Science:

• Medical Imaging
• Computer Vision
• Data Science

Background:

• Accurate segmentation of magnetic resonance (MR) brain images is vital for clinical research and understanding brain tissues.
• Existing segmentation methods face challenges with noise, intensity variations, and inherent data uncertainty.

Purpose of the Study:

• To develop a novel, sustainable approach for segmenting MR brain images using unsupervised clustering.
• To integrate conditional spatial properties and intuitionistic clustering to improve segmentation robustness and accuracy.

Main Methods:

• Proposed a novel intuitionistic-based clustering technique incorporating a hesitation degree to quantify data uncertainty.
• Introduced a conditional spatial function and weighted intuitionistic membership matrix to consider spatial relationships and adapt smoothing.
• Utilized unsupervised clustering for segmenting MR brain images.

Main Results:

• The proposed algorithm demonstrated enhanced robustness with homogenous segments and reduced sensitivity to noise and intensity inhomogeneity.
• Achieved superior performance in retaining image details and segmentation accuracy compared to existing algorithms on synthetic and real datasets.
• Evaluated through qualitative and quantitative parameters including segmentation accuracy, similarity index, true positive ratio, and false positive ratio.

Conclusions:

• The novel intuitionistic-based clustering approach effectively segments MR brain images by addressing data uncertainty and spatial properties.
• This method offers improved robustness, noise resistance, and accuracy, outperforming other algorithms in medical image analysis.
• The technique shows significant potential for applications in the medical industry, enhancing the reliability of brain scan analysis.