Unsupervised Brain MRI Image Segmentation Based on the Finite Mixture of $$\boldsymbol{\alpha}$$-Stable Distributions with EM Algorithm
Ibrahim Sadok, Mourad Zribi, Oussama Fillali · Mathematical Methods of Statistics · 2025
The segmentation of brain magnetic resonance imaging (MRI) plays a crucial role in neuroimaging analysis. Segmentation is the process of converting inhomogeneous data into homogeneous data. Recently, a significant progress has been reported in brain MRI image segmentation. Among these techniques, finite Gaussian mixture models (GMM) are considered to be more recent and accurate. However, GMM is well-suited when the brain MRI image under consideration is symmetric. In reality, medical brain MRI images are often asymmetric. To overcome this problem, we propose an algorithm for brain MRI image segmentation based on a finite mixture of $$\alpha$$ -stable distributions using the Expectation–Maximization (EM) algorithm. The experimentation is conducted with four different brain MRI images, and the results obtained are evaluated using quality metrics. Additionally, this approach is validated through a comparative analysis with CO $${}_{2}$$ emissions data from fossil fuels and industry, measured in tonnes per person, specifically focusing on the significance of utilizing robust statistical models for real-world data applications. This broader context underscores the versatility and importance of the proposed segmentation method.